RuView live sensing in Claude Code: /ruview opens a pane beside the transcript with three views: an overview of your CSI nodes (ESP32, Realtek) and 60 GHz radar (rates, loss, RSSI, presence, device-reported vitals, alerts), a live CSI amplitude waterfall, and a radar range fan with vitals charts, all drawn in terminal cells. Reads only; runs the @ruvnet/ruview CLI it ships beside.
ruflo as a Claude Code mod (function hooks, ADR-404): in-process prompt routing, edit learning signals, tighten-only tool checks from ruflo policy, the cost-tracker budget ladder, a mod trust gate for newly installed mods, and the $.ruflo noun other mods compose their status with. Opt-in and removable: the classic hook-handler hooks stay the default and take every event back whenever the mod is not loaded.
Agent teams, swarm coordination, Monitor streams, and worktree isolation: wraps 4 swarm_* + 8 agent_* MCP tools (12 total) plus 6 topologies (hierarchical / mesh / hierarchical-mesh / ring / star / adaptive). With Claude Code function hooks on (early access), a live swarm pane: agent tiles lit by reads and writes, topology, task and claims board, cost, router pick and hive-mind votes, with buttons that act through the ruflo CLI
Guided modernization for any legacy codebase: start with /modernize, get an assessment, an interactive map, the business rules mined from the code, and a plan you approve, then an upgrade, a rewrite or a rebuild with independent proof that the new code behaves like the old. Includes security hardening, a one-page report, specialist agents, and an optional live progress pane (early access).
Firstmate Calm for Claude Code: the sailboat working animation and conversation-only transcript presentation, sharing the per-home config/calm preference with the Pi Calm extension. Its hooks module may load through CLAUDE_CODE_ENABLE_FUNCTION_HOOKS or Claude Code's tengu_plugin_hooks_modules rollout flag, but the mod activates only when CLAUDE_CODE_ENABLE_FUNCTION_HOOKS is exactly 1 and is otherwise a complete no-op.
Consult multiple AI coding agents (Gemini, OpenAI, Grok, Perplexity, Kimi, any model OpenRouter routes to, plus codex, antigravity, grok, kimi and cursor CLIs when installed, and a local ollama model) to get diverse perspectives on coding problems, with a streaming pane in tmux or inside Claude Code. It can also hand a coding task to a Codex specialist that works on its own git branch and follows skills you pick (experimental).
name: Plugin name "claude-council" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.; ../mods/council-pane/hooks/pane.tsx env writes: COUNCIL_MOD_PANE_DIR; ../mods/council-pane/hooks/pane.tsx env reads: HOME, TMPDIR, TMUX
Jev-powered verbatim compaction: once context reaches 60%, prunes stale tool calls with zero generation. Skips (never falls back to the LLM summarizer) when nothing is prunable; records evidence for learning.db.
Displays lesson hint cards when tool calls match known patterns. Additive surface during pilot; the classic --strict guard keeps its 12 block-severity denies.
Recalls the few memories related to each prompt: one line for the human, one context entry for the model. Local files only, no network on the hot path.
Masks secret values in tool results before the model reads them: named vendor variables, value-shape patterns, and high-entropy tokens. No reveal path.
Suggests (or, opt-in, runs) /compact at a completed checkpoint, judged by TypeSafe Jev. Its hooks module activates only when CLAUDE_CODE_ENABLE_FUNCTION_HOOKS is exactly 1 and is otherwise a complete no-op.
Claude Code plugin: use ChatGPT (Codex), Cursor, OpenCode Zen, and Antigravity models inside one Claude Code session. Switch with /model, delegate to named native workers, keep each provider's own login and permissions.
A Claude Mod for the shunt gateway: /shunt:usage reads the shared account pool's remaining headroom and reset times from shunt's GET /usage, so you can see how close the pool is to throttling without the admin dashboard.
Access & validation details
Observed events
Only what it hooks
Hooks
none
API calls
none recorded
Validation on 2.1.287
fails on 2.1.287
Validator output
modules../register.ts: shunt: /tmp/acm-clones/pleaseai__shunt/plugins/shunt/hooks/register.ts, compiled line 18 `const registration = on("command.run", { command: COMMAND_NAME }, async ($, e, n...`: the value of on("command.run") is kept (assigned, passed, read, or returned from a nested function); it takes .catch(handler) where it is called and nothing else; $ is always spelled $.noun.event(...) at the call site, on is… [+68 chars]
A calibrated context sieve for Claude Code. Every large tool result is judged by a System One model before it enters context; irrelevant blocks are replaced with a summary and a recall key.
Answers Claude Code's built-in WebSearch with Genesis's web_search chain, falling back to the built-in on any failure. Needs CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1.
Guided breathing exercises above the prompt while Claude works: coherent, box, 4-7-8 and the physiological sigh, four animation styles, and the spinner reads the breath. Appears when Claude starts, disappears when Claude answers. Needs function hooks (early access) and an interactive terminal.
name: Plugin name "mindful-claude" reads as one of Anthropic's own. Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.
Watchlist band above the Claude Code prompt: Taiwan hours show the Taiwan list (紅漲綠跌), US hours show the US list (綠漲紅跌), in a broker-style table with a Solari split-flap footer, plus a 損益 mode (P&L view) for your holdings. 20 symbols a market, 10 on screen at a time. Live prices from Yahoo by default, no key and no account; 永豐 Shioaji (macOS/Linux) and 群益 Capital (Windows) are opt-in real-time routes the band runs itself, tried in your own preference order (twSources) - keep that order and any broker paths in a user-level ~/.claude/stock-band.json so a shared project config stays neutral. Needs function hooks (early access) and an interactive terminal. Zero model tokens.
Requests a session compaction when the live context passes a token threshold, so verbatim compaction plugins run near the working ceiling instead of at the model window.
Prints which instruction files the engine loaded (tier, path, @ imports, size) once per context load, and hints when a project CLAUDE.md keeps its AGENTS.md from being read.
Access & validation details
Observed events
the system prompt
Hooks
session.start, prompt.context
API calls
$.env.get, $.fs.stat, $.session.cwd, $.ui.log
Validation on 2.1.287
Passed
Validator output
./register.ts env writes: nothing; ./register.ts env reads: HOME
Scores each unattended turn against the task the person asked for with TypeSafe's Jev model, and after consecutive off-task turns redirects or pauses the next machine-submitted prompt.
cache-tax, the Mod. Everything the hook form does, from inside Claude Code instead of the transcript: stops a cold send once with its price (the resend goes through), keeps the 1-hour prompt cache warm for a window you set with one cache-shared fork per idle stretch, arms that for you after a paid cold write, and keeps score of this session's cold writes. Needs function hooks (early access).
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ./lcm-hooks.ts env writes: nothing; ./lcm-hooks.ts env reads: LCM_SUMMARIZE_WORKER, LCM_SUMMARIZE_WORKER_MAX_OUTPUT_TOKENS, LCM_SUMMARIZE_WORKER_MODEL
Games above the Claude Code prompt (snake, Tetris, 2048, Minesweeper, Flappy, Pong, typing test, Space Invaders, Doom) plus a pet that grows as Claude tests, commits and edits. Clones of the genre, not affiliated with or endorsed by Tetris Holding, Taito, Atari or id Software. Needs function hooks (early access) and an interactive terminal.
Thinking discipline for AI work. Clarify intent, shape approach, consult experts, guard unsupervised work, hand work back so you own it. The slop hook needs CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1.
A grab bag of hunches: experimental skills and plugin features tried here before they earn a place in hope or sound. Everything graduates or dies. The tend mod needs CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1.
Access & validation details
Observed events
classic hooks, every tool call, every prompt, compaction
Pins every subagent to the fable model: rewrites the model on agent.spawn unless the spawn is a fork, with /fable-pin on, off and status. Touches nothing but its own store. Needs function hooks (early access).
Shared Claude runtime adapter: normalises function-hook events into ClaudeRuntimeEvent, exposes $.runtime (emit bus, supports, judge, snapshot) to other plugins, and writes a JSONL event log
name: Plugin name "claude-runtime" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.; types ./types/index.d.ts declares on $: $.runtime; ./index.ts env writes: nothing; ./index.ts env reads: CLAUDE_CONFIG_DIR, HOME, USERPROFILE
CLAUDE RUNTIME X-RAY: wildcard inspector with filters (/xray all|tools|subagents|session|usage|ui|errors|unknown|off), a HUD above the prompt, and a black-box timeline file
COSTCLAW LIVE: per-request cost, cache decay and repeated-read detection while the session runs, and it serves the Nth identical Read of an unchanged file from its own cache instead of paying for it
PRODUCTION GUARD: ambient project/environment context + a transport-independent policy core (DashClaw evidence classifier + offlocal policy engine) enforced natively on tool.call and tool.check. Blocks, asks, or contains before the command runs. /prodguard
SUBAGENT SUPERVISOR: routing-and-budget Mod. Rewrites a spawn's model onto the capability graph instead of denying it, keeps a measured per-agent-type token ledger in $.store, caps tool calls per subagent, and shows live subagents + the last routing decision above the prompt. /supervisor prints the routing log and the ledger.
Diagnoses wrong processes, workflows and habits in Claude Code sessions: fires on plan acceptance, workflow launches, pace complaints and the clock, names the lean alternative with its measured cost, and lets the user fix it in one click.
Access & validation details
Observed events
every tool call, subagent spawns, compaction, every prompt
pi agent for claude: a `pi` subagent type whose model steps run the pi CLI, so pi subagents and agent-team teammates live in Claude Code's own agent UI, streamed live.
name: Plugin name "pi-agent-for-claude" reads as one of Anthropic's own. Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.
runs processes, drives Claude, reads the transcriptValidator warnings
Cheap, fast, typed AI judgments inside Claude Code via the jevctl CLI (verify, screen, find, classify, extract, rerank, match, route, ask, batch), plus verbatim Jev-guided context compaction that replaces the built-in summary.
Shared Claude runtime adapter: normalises function-hook events into ClaudeRuntimeEvent, exposes $.runtime (emit bus, supports, judge, snapshot) to other plugins, and writes a JSONL event log
name: Plugin name "claude-runtime" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.; types ./types/index.d.ts declares on $: $.runtime; ./index.ts env writes: nothing; ./index.ts env reads: CLAUDE_CONFIG_DIR, HOME, USERPROFILE
CLAUDE RUNTIME X-RAY: wildcard inspector with filters (/xray all|tools|subagents|session|usage|ui|errors|unknown|off), a HUD above the prompt, and a black-box timeline file
COSTCLAW LIVE: per-request cost, cache decay and repeated-read detection while the session runs, and it serves the Nth identical Read of an unchanged file from its own cache instead of paying for it
PRODUCTION GUARD: ambient project/environment context + a transport-independent policy core (DashClaw evidence classifier + offlocal policy engine) enforced natively on tool.call and tool.check. Blocks, asks, or contains before the command runs. /prodguard
SUBAGENT SUPERVISOR: routing-and-budget Mod. Rewrites a spawn's model onto the capability graph instead of denying it, keeps a measured per-agent-type token ledger in $.store, caps tool calls per subagent, and shows live subagents + the last routing decision above the prompt. /supervisor prints the routing log and the ledger.
Hands the small judgments in a session to TypeSafe's Jev, a System One model that returns typed answers instead of generating text: what kind of request each prompt is, which model tier a subagent needs, whether an edit breaks a rule already written in your CLAUDE.md or AGENTS.md, and which transcript blocks survive compaction. Kept blocks carry over verbatim: no LLM writes a summary.
name: Plugin name "claude-jev" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.; hooks.UserPromptSubmit: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node --experimental-strip-types ${CLAUDE_PLUGIN_ROOT}/src/prompt-router.ts. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.PreToolUse: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node --experimental-strip-types ${CLAUDE_PLUGIN_ROOT}/src/subagent-router.ts. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.PreToolUse: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node --experimental-strip-types ${CLAUDE_PLUGIN_ROOT}/src/rules.ts. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.PostToolUse: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node --experimental-strip-types ${CLAUDE_PLUGIN_ROOT}/src/rules.ts. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.PostToolUse: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node --experimental-strip-types ${CLAUDE_PLUGIN_ROOT}/src/rules.ts. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.Stop: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node --experimental-strip-types ${CLAUDE_PLUGIN_ROOT}/src/rules.ts. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.PostToolUseFailure: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node --experimental-strip-types ${CLAUDE_PLUGIN_ROOT}/src/rules.ts. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.
runs processes, changes config, persists state, drawsfails on 2.1.287
Records each tool call's permission verdict and wall time so auto-mode classifier cost is measurable from the session index
Access & validation details
Observed events
every tool call
Hooks
tool.check, tool.call
API calls
$.clock.now, $.env.get, $.fs.write, $.session.id
Validation on 2.1.287
Passed with warnings
Validator output
version: No version specified. Consider adding a version following semver (e.g., "1.0.0"); ../mod/register.ts env writes: nothing; ../mod/register.ts env reads: HOME
version: No version specified. Consider adding a version following semver (e.g., "1.0.0"); ../mod/register.ts env writes: nothing; ../mod/register.ts env reads: HERDR_BIN_PATH, HERDR_ENV, HERDR_PANE_ID
awsh for Claude Code: an `aw` subagent type whose model steps are answered by a session on the local awsh harness daemon, so any harness and any backend the daemon can drive runs as a native Claude Code subagent, streamed live. Also carries the Aither World skills, the `aither` output style, MCP servers for installed bricks and portable brick hooks.
Compacts a long tool result through the AitherOS decision door before the model reads it: per LINE SHAPE, learned from outcomes, never dropping a traceback or a summary line.
name: Plugin name "claude-image-generation" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.
runs processes, drives Claude, reads filesfails on 2.1.287
A live workflow pane: replaces the default Workflow progress list with a phase-by-phase graph or a timeline beside the transcript, showing each agent's state, timing, live token spend, tool calls, what it is saying, and prompt.
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ./register.ts env writes: nothing; ./register.ts env reads: HOME
OpenTelemetry for Claude Code plugins: adds $.autotel in the engine.create fold so a plugin records spans, traces every hook dispatch with the chain beneath it, and shows each tool call's settled duration on its row. Exports OTLP/HTTP to OTEL_EXPORTER_OTLP_ENDPOINT (autotel-devtools claude sets it).
Probability-scored guardrails for Claude Code: deny rule-breaking edits and unasked-for deploys, route your own docs into each prompt, and check the final answer against the turn's own evidence.
Access & validation details
Observed events
Edit|Write|Bash calls, Skill calls, every prompt, Agent calls
Compacts when the work moves on from one sub-task to the next instead of when the context window fills, including in the middle of a long turn. Once the context is past a floor, a model judges each step, reading what you asked for and where Claude has got to but never any tool output, for whether the work moves on from a finished piece to another; if it does, taskcut runs Claude Code's own compaction. Below the floor it does nothing and costs nothing.
TypeSafe's Jev decides how Claude Code works on each prompt: the main conversation's reasoning effort (low to xhigh, raised to max mid-turn when tool calls keep failing), each subagent's model, whether to split the work across subagents, and the one skill the prompt needs. Brings in custom OpenRouter models (a junior coder reviewed by Opus, or cheap bulk work) and Codex or OpenCode as reviewers, in modes you choose, each health-checked. Reads the prompt with recent context, records every decision, and suggests its own tuning. Backends: OpenRouter, TypeSafe's API, the Vercel AI Gateway, or Claude Code's built-in classifier. Built on claude-code-templates' jev-model-router and jev-skill-suggestion.
Access & validation details
Observed events
every prompt, every tool call, compaction, the jev-pilot:report skill, subagent spawns, the jev-pilot:setup skill, every skill, /"(?:)"/ calls
The agent flow pane: /flow opens a live tree of the session's subagents and teammates beside the transcript, fed by engine events, with a text fallback where no pane can be drawn.
Web search through Jev Search. Answers Claude Code's built-in WebSearch with Jev (sources and time window chosen by Jev, every result ranked by relevance) and falls back to the built-in search when Jev cannot answer. Also adds the jev_search MCP tool for explicit source or time filters, and a skill on when to use it.
Jev keeps the context window lean so compaction comes later: prunes tool output on entry, caps subagent reports, carries exact facts across compaction, compacts when the cache is cold.
Access & validation details
Observed events
Only what it hooks
Hooks
none
API calls
none recorded
Validation on 2.1.287
fails on 2.1.287
Validator output
modules../verbatim-compact.mjs: HyunjunJeon__jev-context: cannot import "../vendor/jev-pruner/dist/jev.js" (from hooks/verbatim-compact.mjs): no such file under /tmp/acm-clones/HyunjunJeon__jev-context
Client-side secret/PII redaction hooks: keep API keys and sensitive identifiers out of Claude Code's model context by redacting Read/Bash/Grep tool output, blocking prompts that carry secrets, and denying reads of known credential files. Session-transcript scrubbing verified on Claude Code 2.1.263 (updatedToolOutput is what gets persisted); re-verify on other versions.
Plan at the frontiers and review the plan in the browser: decisions and interfaces before mechanics, vertical slices each closed by a check. /vellum:start opens a planning mode of its own: Claude writes the plan and its mockups and grills the reviewer on the open choices; the reviewer comments, edits or approves in a page, and the answer reaches Claude as a prompt.
Access & validation details
Observed events
the vellum:start skill, the vellum:stop skill, mcp__vellum__submit calls, mcp__vellum__state calls, mcp__vellum__grill_ask|mcp__vellum__propose|AskUserQuestion calls, every prompt
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ../src/runtime/hooks/register.ts env writes: CLAUDE_BASH_MAINTAIN_PROJECT_WORKING_DIR; ../src/runtime/hooks/register.ts env reads: CLAUDE_BASH_MAINTAIN_PROJECT_WORKING_DIR
Milestone-driven development workflow and tracking system for Claude Code projects: a language-agnostic core with per-repo toolchain profiles (R, Python, Claude Code plugin, Docker image, or generic), plan/implement/review skills, hotfix and release paths, Fable review briefs, and a self-auditing cairn/ file system.
types ./types/index.d.ts declares on $: nothing (no EngineInterface member); types ./types/index.d.ts declares state: cairn.band; root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ./status/register.tsx state writes: cairn.band; ./status/register.tsx state reads: cairn.band
After an edit that adds a GitHub Actions step pinned to a tag or a branch, names the commit SHA to write instead, from the GitHub API. Needs function hooks (early access).
Picks the recommended option of a question that waited unanswered for a set time, so a session you left does not stop. Off until you turn it on. Needs function hooks (early access).
Shrinks each Bash result before the model reads it: per-command filters for git, test runners, linters, package managers, containers and file tools, the full output kept in a file. Needs function hooks (early access).
Shows the background shell tasks of the session on the status line or in the shared sidebar, with the oldest one's age, and stops one from the /bg-tasks pane. Needs function hooks (early access).
Runs proof-driven bug hunt rounds with /bughunt: each round proves one bug with a failing command the mod runs itself, fixes it, and proves the fix, and the loop stops on a blocked or unverified round. /bughunt collab runs a read-only scanner, planner and critic in sequence. Needs function hooks (early access).
Keeps the 1-hour prompt cache warm for a window you set, or with no end in every session under always, with one cache-shared fork per idle stretch and one keep-warm message after a resume, arms a window for you after a paid cold write, and shows the cache state, the cold price and this session's cold writes. Needs function hooks (early access).
Names the uncommitted files to you at the end of each turn, and tells the model at the next prompt to commit each finished piece as it lands. Needs function hooks (early access).
Parses each JSON, YAML, TOML or .env file an Edit or Write touches, notes a parse error at once instead of at the next build, and says so again when a later edit fixes it. Needs function hooks (early access).
Hands a skill or command call the current text of its file when the file changed on disk after the session loaded it, and hands the model a rules file or the global CLAUDE.md that changed on disk. Needs function hooks (early access).
After the model changes a function signature with Edit, adds the callers ripwire finds to the Edit's result, so the model fixes them before the build does. Needs function hooks (early access).
Asks a council of models about a hard problem, when the model calls its tool or you run /council <question>: Claude models, and Gemini models when gemini-core has a key, answer in parallel, and the model of the session writes one verdict from their answers. Needs function hooks (early access).
Access & validation details
Observed events
classic hooks, the system prompt, /"^mcp__council__convene$"/ calls
Checks each package the model installs against its registry and OSV.dev, and stops a missing, brand-new, look-alike, outdated or vulnerable one, naming the latest version. Needs function hooks (early access).
Renders the mermaid blocks of the model's replies with an installed mmdc after each turn and draws each picture under its reply. Needs function hooks (early access).
Measures the build artifacts of the repository, shows them on the status line from 5 GB, and deletes the ones you pick in the /disk-janitor pane; data directories are never listed. Needs function hooks (early access).
After each commit the model makes, adds the doc lines that commit made stale (file:line references, symbol names) to the commit's result, from ripwire doc-drift, and holds each finding open until the doc holds again. Needs function hooks (early access).
Adds a note to the fifth edit of one file in one turn, so the model re-reads the code path and states the root cause instead of trying again. Needs function hooks (early access).
Has a small model rate how hard each prompt is and runs that turn and its subagents at the matching effort, within the levels you allow, on the models whose prompt cache survives an effort change. Needs function hooks (early access).
After each commit the model makes, names the env variables its added lines read that .env.example lacks, with the file and line of each. Needs function hooks (early access).
Sends one continue prompt after a turn an API error killed, so the half-done work carries on instead of the session going idle, at most 99 times in a row by default. Needs function hooks (early access).
Remembers which tests failed on which code, and tells the model when a failing test has both passed and failed on the same code, so it runs the test again instead of changing code. Needs function hooks (early access).
Gives the model a Gemini advisor tool it calls by itself: Gemini reads the whole conversation and the model's message and answers with a second opinion. Off until /gemini-advisor on; gemini-core holds the key, tier, model and thinking level. Needs function hooks (early access).
Access & validation details
Observed events
classic hooks, the system prompt, mcp__gemini-advisor__advise calls
Moves compaction from Claude to Gemini: in summary mode Gemini summarizes the conversation and the newest messages stay verbatim, in prune mode every message stays and Gemini keeps, cuts or drops older tool calls. Off until /gemini-compact on; gemini-core holds the key, tier, model and thinking level. Needs function hooks (early access).
Keeps the Gemini keys and tier of every Gemini mod, and each mod's model and thinking level, in one place: adds $.gemini, which builds their Gemini requests, reads the answers and moves to the next key after a quota or key error, and picks a model from Google's list. Needs function hooks (early access).
Has Gemini review each plan before the approval dialog, from the plan and the conversation, and sends a plan with a blocking finding back to the model, at most twice. Off until /gemini-plan-review on; gemini-core holds the key, tier, model and thinking level. Needs function hooks (early access).
Has Gemini review every git commit the model runs, from the staged diff and the conversation, and stops a commit with a blocking finding: a bug, data loss, a security hole or a committed secret. Off until /gemini-review on; gemini-core holds the key, tier, model and thinking level. Needs function hooks (early access).
Ships a commit skill and holds every Bash git commit to it: a commit the skill did not open, a blanket git add, a signature trailer, a secret, an ignored path, or a push or branch change nobody asked for is stopped before git runs. Needs function hooks (early access).
Access & validation details
Observed events
every prompt, Skill calls, the git-commit:commit skill, Bash calls
After an edit that calls translation keys, names the keys the JSON, Laravel PHP, YAML or gettext locale files lack, and the languages that lack them. Needs function hooks (early access).
Draws an ASCII animation above the prompt while the model works: matrix rain, fire, an aquarium, a cat, or a GIF of your own imported as a character clip for every project. Display only: nothing reaches the model. Needs function hooks (early access).
Shows the subscription usage limits under the prompt or in the shared sidebar with a reset countdown and a projection, opens a /limit-watch pane, and logs once when a limit passes 80% and 95%. Needs function hooks (early access).
After each commit the model makes, names the manifests whose dependencies it changed without their lockfile (npm, Composer, Cargo, Go, Python, Bundler, Dart, Mix). Needs function hooks (early access).
Tells you when an MCP server failed to connect or dropped, with the engine's reason and a reconnect button, and when it is back; claude.ai connectors are left out. Needs function hooks (early access).
Loads MEMORY.md into the session and saves project learnings to it after every turn through a tool-less fork in the background, without blocking the stop. Needs function hooks (early access).
Names each installed plugin, of every marketplace, whose local clone already offers a newer version, with the claude plugin update command that closes the gap. Needs function hooks (early access).
Lists the servers a Bash call of the model started in this repository and left listening on a port, with their age and session, and stops one with SIGTERM, then SIGKILL. Needs function hooks (early access).
Measures how much context each Bash command spent and tells the model, past a size limit, which narrower command would have answered the same question. Needs function hooks (early access).
Keeps the notes you pin with /pin-note for the whole session and sends them to the model again, word for word, after each compaction and /clear. Off until /pin-note on. Needs function hooks (early access).
For mod development: runs a live check of plugins in a fresh Claude Code session in tmux, types the steps one by one, collects the pane and the transcript, then deletes the temp directory, its transcripts and the inline store files. Needs function hooks (early access).
Learns the short prompts you send often in this project and draws them, with the ones you pin by hand, above the prompt; with the prompt box empty, a digit key sends one at once. Needs function hooks (early access).
Writes a prompt longer than the limit (2000 characters by default) to a temp file and sends the model its first 200 characters with the path, so one paste does not fill the context. Needs function hooks (early access).
Draws the time under each of your messages and each text block of the model's replies, also for a resumed session whose transcript is 4 MiB or smaller. Display only: nothing reaches the model. Needs function hooks (early access).
Keeps project and global memories in SQLite through a shared local daemon, hands the model the entries that match its file tool calls, prompts and subagent tasks, and saves new ones with a haiku consolidator after each main-loop turn. Needs function hooks (early access).
Access & validation details
Observed events
/"^mcp__sage-memory__"/ calls, the system prompt, classic hooks, compaction, /"^(Read|Grep|Glob|LSP|Edit|Write|NotebookEdit|MultiEdit|mcp__(?!sage-memory__).+)$"/ calls, every prompt, subagent spawns, /"^(Edit|Write|NotebookEdit|MultiEdit)$"/ calls, Bash calls
For mod development: lets the model run a slash command in its own session, such as /reload-plugins after a plugin update, and read the output as the next prompt. Needs function hooks (early access).
Shows this session's state in the sidebar: context fill, token totals, cost, model and effort, the Claude Code version, the other sessions of this machine, and the git branch and status. Needs function hooks (early access).
Draws each PNG or JPG the model saves or reads under its tool row: pixels in kitty and Ghostty, quadrant block cells in every other terminal. Needs function hooks (early access).
Opens one shared pane beside the transcript and draws the sections, lines and buttons every other mod writes into it through $.sidebar; each stream entry is stamped with its time and kept in a per-project log the next session takes back. Needs function hooks (early access).
Runs slash commands joined with && one after another, waits for each command's turn, pane or dialog to end, and stops the chain at a command that failed or a turn that ended without an answer. Needs function hooks (early access).
Access & validation details
Observed events
mcp__slash-chain__fail calls, every skill, every prompt
After an edit that builds SQL by joining or interpolating strings, names each line, so the model passes the values as query parameters. Needs function hooks (early access).
After an edit that keeps browser data in localStorage or sessionStorage, names each line, so the model stores the data in a cookie instead. Needs function hooks (early access).
Shows each subagent of the session with its turns, time, model and tokens, the costliest first, yellow while it runs and green once it answered, and marks one that passed the token limit. Needs function hooks (early access).
Submits a continue prompt while the task list has unfinished tasks, at most 99 times in a row by default, and turns the task tools on for every model unless CLAUDE_CODE_ENABLE_TODO_TOOLS is set. Needs function hooks (early access).
Stops the model from repeating a tool call that just failed with the same input, until a file or command has changed something. Needs function hooks (early access).
After a curl or wget an automated-client filter answered 403 or 429, gives the model the browser User-Agent to retry with, and the two cases where it must not. Needs function hooks (early access).
An output style that classifies each message by exchange type and shapes the reply to fit it: the answer first, a ceiling per type, reference codes that survive the session, and decisions handed back as questions
The client for ultrapowers: /ultrapowers <plan-path> commits an approved plan and runs on an exe.dev fleet you provision: a disposable sandbox runs the plan as a pool of tasks, a pool of builders, the plan's own probes and checks as the proof, and opens its own PR. No local engine.
Three-level decision layer on TypeSafe's Jev: picks model and effort per session and per subagent, compacts context without rewriting it, and gates code review with calibrated closed questions. Pairs with lexi: the review runs after the last GREEN.
Compacts the main conversation once it has sat idle for idleMinutes, while the prompt cache is still warm, and hands every compaction of the main conversation (manual, auto, plugin, precompute) the Compact Instructions section of the user's AGENTS.md. Skips while a turn runs, below minMessageTokens of conversation, and past cutoffMinutes (a wake from sleep, when the cache has likely expired), saying which in the transcript; a reload of the plugin keeps the pending idle stretch.
Route work through the smallest sufficient loop: handle bounded tasks directly, fan out clear branches, or coordinate durable app-rooted Claude and Codex workrooms with provider-neutral status and git lifecycle tracking.
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ./jev-pruning.ts env writes: STRAW_BOSS_JEV_READY_SESSION, STRAW_BOSS_JEV_WAITING_USAGE; ./jev-pruning.ts env reads: STRAW_BOSS_JEV, TYPESAFE_API_KEY
Specialists execution runtime in Claude Code: specialist activation and dispatch over MCP, session continuity hooks, and the supervising-activations skill. Consumes the Substrate work authority; it is not Substrate.
Stops a session before it runs past a budget. Three figures are guarded, each with a limit you set: the session cost in dollars, the 5-hour plan window and the 7-day plan window. It says nothing until a figure nears its limit, then pins a status line and warns once; past the limit it refuses tool calls and asks before sending another prompt. /guard shows the state and changes the limits.
A live /context: a pinned line under the prompt with the window fill and the growth per turn, and /context-lens for a pane with the per-category bars and how many turns are left before auto-compaction.
Shows the subscription plan limits - the 5-hour window, the 7-day window and a gateway spend limit - as a pinned status line under the prompt, with a reset countdown and a projection of when you would hit 100% at the pace of the last few samples. /quota opens a pane with a bar, the reset clock time and the burn rate for each window.
A per-turn cost and token ledger: a pinned status line with the session cost, the last turn's tokens and the cache hit ratio, and /ledger for a table of the recent turns with their model, tokens, cache reads and writes, duration and cost.
Draws the agentic-workflows engine's gates in the band above the prompt: rows a click puts in the prompt and a second click sends: instead of a menu the model has to reproduce.
LLM-maintained knowledge base for Obsidian vaults. Auto-triggers on paper sharing, vault health queries, concept compilation, research questions, and domain management.
Access & validation details
Observed events
Bash calls, Write|Edit calls, every tool call, subagent spawns, every skill, the system prompt, every prompt
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; hooks.SessionStart: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node ${CLAUDE_PLUGIN_ROOT}/bin/commonplace cleanup-cache 2>/dev/null || true. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.UserPromptSubmit: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node ${CLAUDE_PLUGIN_ROOT}/bin/commonplace prompt-context. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.PreToolUse: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node ${CLAUDE_PLUGIN_ROOT}/bin/commonplace agent-guard. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.PostToolUse: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: node ${CLAUDE_PLUGIN_ROOT}/bin/commonplace post-write. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.
Keeps secrets, email addresses and IP addresses out of the session transcript: swaps each one for a stable placeholder on the way in, and puts the real value back on the way into a tool call
Consultation and prompt engine for Claude Mods. The module is the core; [probe.<id>] tables in probes.toml are the consultants and [prompt.<id>] tables author a system-prompt section, a tool description or a command description. The only carrier is the mod: a CLAUDE_*_CARRIER handle set to mod or left empty/unset arms the probe here; any other value loudly refuses the calls this probe acts on (the patch-side copy is gone).
Access & validation details
Observed events
the system prompt, every tool call, classic hooks, subagent spawns
Multi-line status HUD above the prompt over data families (hooks/data/*.ts pure collectors, one core subscription per event), with the /statusline-mod pill panel: live preview, six tabs (layout, elements, per-element settings, view axes, themes, number formats). Based on konsta95/ClaudeCodeMods statusline 0.3.2 (MIT, c49ad4a).
Routes each Claude Code turn to the model and effort TypeSafe's Jev picks, weighed against what a switch costs, and compacts with Jev instead of a lossy summary.
name: Plugin name "jev-claude-router" reads as one of Anthropic's own. Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.; ./register.ts env writes: nothing; ./register.ts env reads: AI_GATEWAY_API_KEY, JEV_ROUTER_ALLOW_CUSTOM_BASE, JEV_ROUTER_ALLOW_OVERRIDE, JEV_ROUTER_CACHE_TTL, JEV_ROUTER_CEILING, JEV_ROUTER_COMPACT, JEV_ROUTER_COMPACT_MIN_REDUCTION, JEV_ROUTER_COMPACT_TIMEOUT_MS, JEV_ROUTER_EXCLUDE, JEV_ROUTER_JEV_MODEL, JEV_ROUTER_NOTIFY_CONTINUE, JEV_ROUTER_PRICE_CHECK, JEV_ROUTER_PROVIDER, JEV_ROUTER_STICKY, JEV_ROUTER_STICKY_CONFIDENCE, JEV_ROUTER_TIMEOUT_MS, JEV_ROUTER_UPGRADE_MAX, TYPESAFE_API_KEY, TYPESAFE_BASE_URL
Draws the agentic-workflows engine's gates in the band above the prompt: rows a click puts in the prompt and a second click sends: instead of a menu the model has to reproduce.
[Experimental] Upcoming Google Calendar events in-session: /cal toggles a band above the prompt with the next 7 days, /cal N for N days (Claude Mod; needs CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1 and the claude.ai Google Calendar connector)
Obsidian Workspace: project-scoped vault productivity (capture / notes / PM). Skills own folder layout and templates; /obw:pm splits a spec into blocking tickets; /issue is a Claude Mod pane that lists a view of the project's dashboard, draws Mermaid as text diagrams, and can open a card in the browser through viz. Vault I/O uses the Obsidian CLI via the official obsidian-cli skill.
[Experimental] Every omp provider's remaining quota in-session: a /quota band above the prompt, /quota refresh, and a toast when a provider worsens from ok (Claude Mod; needs CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1)
Guided modernization for any legacy codebase: start with /modernize, get an assessment, an interactive map, the business rules mined from the code, and a plan you approve, then an upgrade, a rewrite or a rebuild with independent proof that the new code behaves like the old. Includes security hardening, a one-page report, specialist agents, and an optional live progress pane (early access).
Bullswarm as a Claude Mod: live subscription meters for every installed coding-agent CLI above the prompt, quota-aware routing of Claude's own subagents to whichever pool has surplus, a verdict note on every bullswarm command the model runs, a /bullswarm command, and $.bullswarm for other plugins.
Access & validation details
Observed events
the system prompt, Agent calls, Bash calls, /"^mcp__bullswarm__route$"/ calls
extensions: Unknown field 'extensions'. Claude Code ignores it at load time.; ./fast-jev.ts env writes: nothing; ./fast-jev.ts env reads: TYPESAFE_API_KEY
Mermaid diagrams in Claude Code: every ```mermaid block Claude writes is drawn as box art, in colour, right where the fence was in the transcript. Needs function hooks (early access) and an interactive terminal.
name: Plugin name "claude-mermaid" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.
A prompt typed while Claude is working is held in a stack above the prompt instead of landing in the running turn, and sent once the turn ends. Reorder the stack, edit a row in place, send one first, remove one, flush the lot. Needs function hooks (early access) and an interactive terminal.
name: Plugin name "claude-queue" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.
drives Claude, writes the prompt box, drawsfails on 2.1.287
Every tool call your agent makes, checked before it runs. Verifies each pending tool call against the session's plan using TypeSafe AI's Jev model, then allows it, asks you, or blocks it.
Retell what Claude just said, in plain words: in the band above the prompt, or in a split pane when running under Herdr. Nothing enters the transcript or the model's context.
The T-rex from the browser's offline page, above the Claude Code prompt: jump cacti and duck birds while Claude works. Needs function hooks (early access) and an interactive terminal.
Jev-scored context pruning for Claude Code through OpenRouter: at compaction every tool call and result is scored by TypeSafe's Jev decision model, stale ones are dropped or truncated, everything kept stays verbatim. An optional gate truncates oversized tool outputs as they arrive.
Picks the model and the reasoning effort for each task with TypeSafe's Jev, a System One decision model reached either through TypeSafe's own API or the Vercel AI Gateway. Routes subagent models at agent.spawn and the main loop's effort and model at turn.step, in both directions, falling back to the engine's built-in classifier when no key is set.
Firstmate Calm for Claude Code: the sailboat working animation and conversation-only transcript presentation, sharing the per-home config/calm preference with the Pi Calm extension. Its hooks module may load through CLAUDE_CODE_ENABLE_FUNCTION_HOOKS or Claude Code's tengu_plugin_hooks_modules rollout flag, but the mod activates only when CLAUDE_CODE_ENABLE_FUNCTION_HOOKS is exactly 1 and is otherwise a complete no-op.
/pacman: Pac-Man played over the conversation on screen, eating it character by character while Claude works; Esc puts it back. Needs function hooks (CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1).
The Chrome offline dinosaur, above the Claude Code prompt. Jump the cacti and duck the pterodactyls while Claude works. Needs function hooks (early access) and an interactive terminal.
A pane beside the transcript with the GitHub pull requests and issues related to the current session: drag a title or description to leave a comment on it, add as many as you like, then Submit sends them all as one prompt. Watch each pull request's checks, review decision and mergeability update on a timer.
Gives a Claude Code session a character persona, with a time-aware liveliness layer and a pixel portrait panel; bundles three characters from claudecafe.dev.
Access & validation details
Observed events
Only what it hooks
Hooks
none
API calls
none recorded
Validation on 2.1.287
fails on 2.1.287
Validator output
modules../function/register.js: persona-panel: cannot import "../../../character-core/src/index.ts" (from hooks/function/register.js): it is outside the plugin's folder (/tmp/acm-clones/minipai__claudecafe/packages/persona-panel)
Arcade games above the Claude Code prompt (a dodge race, Breakout, a dino runner and a side-scrolling shooter) that pause when Claude finishes a turn and react to the edits, tests and commits Claude makes. Needs function hooks (early access) and an interactive terminal.
name: Plugin name "claude-games" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.
Pins the model that background recon subagents run on. An agent.spawn hook sets the Explore subagent's model to haiku, so a session driven by an expensive model stops billing its codebase searches at that tier, without shadowing the built-in agent definition and without the CLAUDE.md load a replacement definition would cost.
The review gate as a tool.check function hook. Every shell call is decided in process, so the non-commit calls -- almost all of them -- cost no subprocess at all, and a blocked commit can be put to you as a question instead of refused outright. The decision itself is still bin/review-gate, run only for a command that mentions a commit, so both routes read the same receipts and cannot disagree.
/linear: your Linear projects, milestones and issues in a pane beside the transcript. Read an issue, then press plan, execute or product to send a ready prompt about it into this session. A Claude Code function-hooks (Mods) plugin.
A status line under the prompt: context, cache hit rate, cache TTL countdown, today's spend for this project, and why a turn missed the cache. /usage prints the project ledger.
Runs long Claude Code jobs overnight within your usage limits: pauses just before the 5-hour or weekly limit, resumes the same session after the reset (or relaunches it days later) and works through a list of jobs you hand it with /noctis:start <file>. Also runs in Codex CLI, Antigravity CLI, Droid and Copilot CLI.
Keeps a per-turn tally of edits, runs, curls and destructive commands and shows it on the turn footer; any non-read-only Bash counts as a run, a claim (English or Japanese) with no run flags the spinner, a destructive Bash unfolds its tool group, and an edit-without-run turn suggests running the tests.
version: No version specified. Consider adding a version following semver (e.g., "1.0.0"); ./register.ts env writes: nothing; ./register.ts env reads: CLAUDE_MODS_DISABLE
version: No version specified. Consider adding a version following semver (e.g., "1.0.0"); ./register.ts env writes: nothing; ./register.ts env reads: EDITOR, PATH, REVDIFF_POPUP_HEIGHT, REVDIFF_POPUP_WIDTH, VISUAL, ZELLIJ
Choose subagent models with /agent-models and click the prompt panel to inspect activity, token usage, and routing. Requires Claude Code Mod function hooks enabled.
Shapes and reorders Persian, Arabic and Hebrew text in the transcript through fribidi so the terminal draws it joined, right-to-left and right-aligned.
Keeps the skill listing out of the context window and lets TypeSafe's Jev, a System One decision model, pick at most one skill per prompt from the skills' descriptions, the way TypeSafe's skill-suggestion cookbook does: one request ranks every skill and asks whether the prompt needs a skill at all, a second re-reads the top three with their SKILL.md and can reject them all. The winner's SKILL.md is attached to the prompt, so the skills can be hidden from the model altogether (/jev-skill-suggestion:setup makes them user-invocable-only and /context counts them at 0). Falls back to the engine's built-in classifier when no key is set.
Access & validation details
Observed events
every prompt, compaction, the jev-skill-suggestion:setup skill, every skill
Runs the Laya decision model on your machine: checks each prompt, hints Claude at the right folder and task type, gives Claude a fast laya decide tool, and shows every decision in a band above the prompt.
Are you moving or hiding in the dark? Watches the session for shipped work (commits, pushes, deploys, passing tests) against edits to tooling, config and docs, and says so in one quiet line when you have been in the dark too long. Zero model tokens, no network, nothing leaves your machine.
Gates git work: no commit, push or merge without the user's say-so in their latest message; a push to a branch that already landed is refused; a pushed series must leave the project working at every commit; commit messages (format and the issue it is about, then a Haiku review of the body) and MR descriptions kept tidy.
Access & validation details
Observed events
every prompt, /"^mcp__(plugin_)?git-gates__grant$"/ calls, Bash calls
Docs worth keeping: docs-review asks Haiku whether new text will be read again, docs-no-repeat-code blocks lines that restate code, limit-docs sends a follow-up when a turn adds too much prose.
Keeps the last part of your 5-hour and weekly quota for you. At the reserve, spare10 holds all work and asks you whether to continue. After a Resume, it asks again at the floor, by default at 5% left. Shortly before the reset, it lets work use the reserve.
Access & validation details
Observed events
every tool call, /"^AskUserQuestion$"/ calls, every prompt
Firstmate Calm for Claude Code: the sailboat working animation and conversation-only transcript presentation, sharing the per-home config/calm preference with the Pi Calm extension. Its hooks module may load through CLAUDE_CODE_ENABLE_FUNCTION_HOOKS or Claude Code's tengu_plugin_hooks_modules rollout flag, but the mod activates only when CLAUDE_CODE_ENABLE_FUNCTION_HOOKS is exactly 1 and is otherwise a complete no-op.
Catch wild creatures above the Claude Code prompt: where your prompt leads (water, grass, fire, bugs, plans, builds) decides what appears. Pick a partner that draws under the logo, beside the spinner and above the prompt, and levels up and evolves as Claude works. Needs function hooks (early access) and an interactive terminal.
A read-only side pane beside the transcript: it opens the file Claude last edited at the end of the turn, lists directories one page at a time, draws code highlighted and Markdown rendered, and sends a selected passage back to Claude with a command or the next prompt.
A live checklist above the prompt of the conversation's loose ends: questions left unanswered, points made in passing, caveats and deferred work it moved past, kept up to date by a small model after each turn. Needs function hooks (early access) and an interactive terminal.
Pi Outliner integration for Claude Code: typed outline, workboard and door tools for agents (read, find, edit, comment, changes, draft.patch, ep0ch act), clickable references, and completed responses that mention [[pages]], ((block ids)) or Work IDs in the Outliner's Recent Mentions navigator.
Access & validation details
Observed events
the system prompt, ? calls, mcp__pi-outliner__show calls
Context compaction that prunes instead of summarising: rules drop exploration output and hook noise locally, Jev ranks the rest against a budget, a note says what was removed, and removed outputs are archived for the brain. Falls back to the built-in summary when it cannot do better.
Companion to jev-compact: watches context use after every turn and starts a compaction at the configured percentage. Separate plugin on purpose: the engine skips a plugin's own session.compact hook when that plugin raises the compaction, so the trigger must come from outside jev-compact for the pruning to run.
A Beadwork ticket pane inside Claude Code: /bw <id> or a search box runs bw show and draws the digest, title, description and comments; every ticket id a reply mentions gets a button that opens it. Needs function hooks (CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1).
Read-only side chat: /aside opens a pane beside the transcript where you ask about the session so far; answers come from a tool-less fork of the session's own transcript and nothing is written back into the main thread.
types ./types/index.d.ts declares on $: nothing (no EngineInterface member); types ./types/index.d.ts declares state: time-check.pending, time-check.isSessionMuted, time-check.feedback; root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ./register.tsx state writes: time-check.feedback, time-check.isSessionMuted, time-check.pending; ./register.tsx state reads: time-check.feedback, time-check.isSessionMuted, time-check.pending
Before each prompt reaches the model, TypeSafe's Jev ranks your installed skills against it and attaches one line naming the skill to load. Sends the prompt text to api.typesafe.ai. Needs function hooks (early access).
A Claude Mod: /crosstalk opens a pane with this session's conversations with your other Claude Code sessions (peer deliveries in, SendMessage out) and lets you reply from it.
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ./register.ts env writes: nothing; ./register.ts env reads: CLAUDE_CONFIG_DIR, HOME
Background reflection on Claude Code sessions that promotes recurring lessons into skills, hooks, rules and agents, tracks artifact usage feedback, and ships a local web UI to review and rate them. Runs on Bun, no Python required.
Access & validation details
Observed events
Only what it hooks
Hooks
none
API calls
none recorded
Validation on 2.1.287
fails on 2.1.287
Validator output
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; modules.../dist/hook-module.js: kolezka__self-improvement-loop: /tmp/acm-clones/kolezka__self-improvement-loop/dist/hook-module.js: no such file
ClaudeCodeStatusline as a function-hook mod: the status bar drawn in the hint line under the prompt, each segment revealing its details on hover, with /statusline-mod to pick the segments.
hooks.SessionStart: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: ${CLAUDE_PLUGIN_ROOT}/hooks/check-cli-version.sh. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.SessionStart: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: ${CLAUDE_PLUGIN_ROOT}/hooks/session-baseline.sh. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.SessionStart: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: ${CLAUDE_PLUGIN_ROOT}/hooks/ensure-agent-teams.sh. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.Stop: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: ${CLAUDE_PLUGIN_ROOT}/hooks/suggest-simplify.sh. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.; hooks.Stop: Shell command uses ${CLAUDE_PLUGIN_ROOT} without quotes: ${CLAUDE_PLUGIN_ROOT}/hooks/push-check.sh. If the expanded path contains a space the command can split into several words and fail. Wrap the placeholder in double quotes, or use exec form: {"command": "<executable>", "args": ["${CLAUDE_PLUGIN_ROOT}/..."]}.
Discover Agent Skills under .agents/skills (project) and ~/.agents/skills (user), expose them as slash commands, list them in prompt.context, and optionally intercept the Skill tool so names from those trees load without living under .claude/skills.
Access & validation details
Observed events
Only what it hooks
Hooks
none
API calls
none recorded
Validation on 2.1.287
fails on 2.1.287
Validator output
modules../register.ts: Leechael__claude-code-mod-agent-skills: /tmp/acm-clones/Leechael__claude-code-mod-agent-skills/hooks/discover/scan.ts does not parse: Expected ";" but found "$" (line 89, column 50); Invalid flag "k" in regular expression (line 89, column 68); Invalid flag "l" in regular expression (line 89, column 70); Invalid flag "l" in regular expression (line 89, column 71); Duplicate flag "s" in regular expr… [+27 chars]
An orange-and-white cat chases a butterfly across seven rows above the prompt for as long as Claude works, bounding after it, leaping when it dips low, and going over in a heap when it misses; the trees and grass of the lot slide past behind it as it runs. Needs function hooks (early access) and an interactive terminal.
name: Plugin name "claude-maru-run" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.
Marks conversation content wrapped in <pin>...</pin> tags as compaction-proof: pinned content is re-injected verbatim into the transcript after every compaction (manual /compact, auto-compaction, plugin-triggered, or precompute).
Keep unrelated topics from pulling Claude off your code: listed terms become stable placeholder names (or their lines are dropped) in tool output, prompts, CLAUDE.md and skills before the model reads them. The system prompt is never changed.
Access & validation details
Observed events
every tool call, every prompt, the system prompt, every skill
Plays Doom-engine WADs in a Claude Code pane: runs doom-cli (GPLv2, built from upstream) in a private tmux server and repaints its frames into a Raster. Fetches Freedoom (BSD) on demand. DOOM is a ZeniMax trademark; not affiliated with id Software, ZeniMax or Anthropic.
Required workflow checks with Claude Code function hooks. Refuses a knowledge-file write, or a memory-service save in external memory mode, until knowledge-save is open, refuses hand edits to generated indexes, refuses a pull request, close or merge until knowledge-save (and the work or merge-and-clean-up skill) is open, and holds a final reply once when working memory or the index rebuild and checker did not follow the change. Fact checks only: no model call and no reading of the agent's words. Runs only when CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1.
Access & validation details
Observed events
compaction, every prompt, classic hooks, every skill, every tool call
Asks what to do before Claude Code re-sends a large conversation whose prompt cache has expired: on a cold resume, and on the first prompt after an idle spell.
A pane beside the transcript that holds the next things to tell the agent: write blocks of text while a turn runs, then press [+] to put one block into the prompt box.
A Claude Mod for the shunt gateway: /shunt:usage reads the shared account pool's remaining headroom and reset times from shunt's GET /usage, so you can see how close the pool is to throttling without the admin dashboard.
Access & validation details
Observed events
Only what it hooks
Hooks
none
API calls
none recorded
Validation on 2.1.287
fails on 2.1.287
Validator output
modules../register.ts: shunt: /tmp/acm-clones/miguelrisero__better-shunt/plugins/shunt/hooks/register.ts, compiled line 18 `const registration = on("command.run", { command: COMMAND_NAME }, async ($, e, n...`: the value of on("command.run") is kept (assigned, passed, read, or returned from a nested function); it takes .catch(handler) where it is called and nothing else; $ is always spelled $.noun.event(...) at the call … [+79 chars]
Guided modernization for any legacy codebase: start with /modernize, get an assessment, an interactive map, the business rules mined from the code, and a plan you approve, then an upgrade, a rewrite or a rebuild with independent proof that the new code behaves like the old. Includes security hardening, a one-page report, specialist agents, and an optional live progress pane (early access).
Lossless compaction for Claude Code: nothing is summarized; everything removed is archived verbatim with provenance and restorable by id (Jev or a local ruleset). /lossless to inspect and restore.
A Claude Mod (function-hooks plugin): /aquarium draws the session as a fish tank in a pane with kitty graphics: the main loop is the parent fish, each subagent a baby that is born and swims off, and every tool call an animation (Read feeds, Edit nests, Bash dashes, an error sinks); the water level follows the context window. Half-block fallback where the terminal shows no images. / Claude Mod（function hooks プラグイン）。/aquarium でセッションを Kitty graphics の水槽として pane に描画。メインループが親魚、サブエージェントは誕生して泳ぎ去る稚魚、ツール呼び出しは魚の動き（Read で餌、Edit で巣、Bash でダッシュ、エラーで沈む）、水位はコンテキスト使用率に連動。画像非対応端末では半ブロック描画に自動切替。
A Claude Mod (function-hooks plugin): /git draws a clickable panel above the prompt for everyday git: pick a branch with a button (or filter + Enter, or a digit key), confirm, and it checks out; pull/push/fetch, stage-all and commit are buttons too. No force-push, reset or rebase exposed. / Claude Mod（function hooks プラグイン）。/git でプロンプトの上にクリック操作できる git パネルを表示。ブランチをボタン（絞り込み+Enter、数字キーも可）で選び、確認してからチェックアウト。pull/push/fetch、stage all、commit もボタンで実行。force-push・reset・rebase は意図的に非対応。
Paints Markdown onto Claude Code's prompt box as you type: fenced code becomes a syntax-highlighted card (even before you close the fence), and inline code, bold, italic, lists, task lists, quotes, tables, links and headings are styled. Paint only, your text is never changed.
Automated multi-agent code review cycle. Spawns reviewers in parallel (plus Codex when available), applies fixes per CLAUDE.md policy, admits a commit only when a reviewer saw exactly what it records, and asks before a push you did not request.
Typed judgement calls for Claude Code: two primitives, judge and rank, under packs that grade issues, pull requests, plans, commits, releases and rules, a repository watch, tool output pruning and an outbound text gate, backed by TypeSafe Jev or the session's small model.
Verbatim Jev-guided compaction for Claude Code sessions.
Access & validation details
Observed events
Only what it hooks
Hooks
none
API calls
none recorded
Validation on 2.1.287
fails on 2.1.287
Validator output
modules../fast-jev.ts: fast-jev-compaction: /tmp/acm-clones/PAVincius__cosmos-nebuloz/.maestri/plugins/fast-jev-compaction/hooks/fast-jev.ts, compiled line 147 `const fromEnv = await $.env.get(name);`: $.env.get takes a literal name as its first argument, so the variables a module reads and writes can be listed (got the variable name); $ is always spelled $.noun.event(...) at the call site, on is always on("<event>", ho… [+44 chars]
The Loadout's core work board and orchestration loop for Claude Code. Tickets are classified into categories that route each one to a concrete model and effort, then dispatched to token-gated executors; side issues you mention mid-task get captured on the spot, and a live, self-hosted Kanban dashboard spans every project you work in.
Replaces context compaction in the controller's interactive session of a log-driven repo: the conversation is rebuilt from .context/events.jsonl by `eventlog context`, plus the last few turns word for word. Rebuilds at the log's task boundaries and at a fill backstop; /rebuild queues a rebuild for the end of the next turn, and /compact rebuilds from the log at once. Inert in headless sessions, for other EVENTLOG_AS writers, with EVENTLOG_CONTEXT=off, and in a repo without the log. Falls back to the engine's compaction when eventlog fails.
Cleans up finished git worktrees as soon as a turn or a session ends. When a turn ends (the main session's or a subagent's), it runs `offcut apply --only auto --yes` for the session's repo in the background: a worktree goes only if it is clean, pushed, and its work is merged or eventlog says its agent is done. When the session ends, it runs once more a few seconds later, also clearing unused build caches, so the session's own worktree can go. Every removal is recorded for `offcut restore`. Inert when `offcut` is not installed, outside a git repo, or with OFFCUT=off.
name: Plugin name "claude-who" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.
network, runs processes, persists state, drawsfails on 2.1.287
version: No version specified. Consider adding a version following semver (e.g., "1.0.0"); types ./hooks/agentic-plugin.d.ts declares on $: nothing (no EngineInterface member); types ./hooks/agentic-plugin.d.ts declares state: agentic-plugin.memqLaunchDir; root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ./index.ts env writes: nothing; ./index.ts env reads: HOME, TYPESAFE_API_KEY, USERPROFILE; ./index.ts state writes: agentic-plugin.memqLaunchDir; ./index.ts state reads: agentic-plugin.memqLaunchDir
Active Directory administration over LDAPS for Claude Code: an MCP server (FastMCP + ldap3) with 15 tools, admin agents, slash commands, a safety skill, and a guard that refuses any AD write until the identical call has been dry-run first. Headline workflow: mass-assign computer managedBy from description-held owner names via a dry-run plan/review/apply loop.
Access & validation details
Observed events
Only what it hooks
Hooks
session.start, session.end, tool.check {tool=?}
API calls
$.agent.list, $.session.messages
Validation on 2.1.287
Passed with warnings
Validator output
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.
Routes each turn and subagent to a model tier and reasoning effort using a classifier behind any server that speaks POST /v1/classify (model-router-api on Cloudflare Workers AI today), falling back to the engine's built-in classifier.
version: No version specified. Consider adding a version following semver (e.g., "1.0.0"); ./model-router.ts env writes: nothing; ./model-router.ts env reads: HOME
Firstmate Calm for Claude Code: the sailboat working animation and conversation-only transcript presentation, sharing the per-home config/calm preference with the Pi Calm extension. Its hooks module may load through CLAUDE_CODE_ENABLE_FUNCTION_HOOKS or Claude Code's tengu_plugin_hooks_modules rollout flag, but the mod activates only when CLAUDE_CODE_ENABLE_FUNCTION_HOOKS is exactly 1 and is otherwise a complete no-op.
A private text chat with another human while Claude works, with random matching, interests, next, block, report, and automatic task completion handling.
name: Plugin name "claude-roulette" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.; ./register.ts env writes: nothing; ./register.ts env reads: ROULETTE_URL
network, reads env vars, persists state, drawsfails on 2.1.287
Corrects the model tier of every subagent spawn (agent.spawn) with a live Jev judgment (haiku / sonnet / opus), instead of relying on /delegate's static routing table. A risky task (touches production, money, or irreversible data) is always forced to opus regardless of confidence.
An unofficial fan-made pixel cat that lives in the band above the prompt and reacts to what Claude is doing: thinking, editing, craning after a search, yowling in a shell, squeezing its eyes shut at a tool error, and folding its ears down to sleep between turns.
Picks the reasoning effort for each turn of the main conversation with a System One classifier (hosted Jev or a compatible service) and shows the level on the spinner and on the line that closes the turn. In shadow mode it shows and logs its pick and sends the session's effort; in enforce mode it sends the pick. Requires CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1.
Harness engineering for Claude Code: attribute every prompt/agent change, keep every candidate, diagnose from traces, hold out a split - and search over Claude Code's own scaffold against your repo's real failures.
Brings back /buddy: the small creature that watched you code, above the prompt again. Restores the companion in ~/.claude.json exactly (same species, rarity, hat and stats), hatches an egg for new accounts, levels up from commits and passing tests, unlocks a 49-buddy roster as you ship, reacts to Claude's work for free, answers in a Haiku-written line when you say its name, and shows in a pane on desktop and mobile. Needs CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1.
Replaces the spinner with an animated ASCII slot machine: pull the lever on every prompt, watch the reels spin while Claude works, cash in when the turn completes.
Access & validation details
Observed events
Only what it hooks
Hooks
none
API calls
none recorded
Validation on 2.1.287
fails on 2.1.287
Validator output
name: Plugin name "claude-slots" is reserved: it passes as one of Anthropic's own. A third party's plugin name cannot start with "claude-", "anthropic-", "anthropics-", or "cc-plugin-", be "claude", "anthropic", "anthropics", "claude-code", or "claude-mods", or put "official" beside "claude" or "anthropic". Name it for what it does. If this is one of Anthropic's own plugins, validate the marketplace that lists it.; modules../slots.js: WorldInnovationsDepartment__claude_slots: /tmp/acm-clones/WorldInnovationsDepartment__claude_slots/hooks/slots.js:407: the hook on "turn.step" is not an async generator: turn.step streams, so it takes async function* ($, e, next) { ... }, which yields the chunks and returns the result; $ is always spelled $.noun.event(...) at the call site, on is always on("<event>", hook), and next.to always next… [+16 chars]
Guided modernization for any legacy codebase: start with /modernize, get an assessment, an interactive map, the business rules mined from the code, and a plan you approve, then an upgrade, a rewrite or a rebuild with independent proof that the new code behaves like the old. Includes security hardening, a one-page report, specialist agents, and an optional live progress pane (early access).
A state-machine graph engine that drives a ticket through claim -> worktree -> coding -> evaluate -> review -> human approval, spawning and monitoring subagents at each step.
/ale-board: a live task board for an Agentic Label Engineering (ALE) run above the prompt and in a docked pane. Reads `python3 -m ale status --json` and the run's label files; read-only, LLM-free. An experiment on Claude Code function hooks (Mods).
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ../mod/register.tsx env writes: nothing; ../mod/register.tsx env reads: ALE_ROSTER, ALE_RUN_DIR
/music: what macOS Music.app is playing, in one line above the prompt. A pressable play/pause glyph, track · artist · album, a progress bar with the clocks, and a pressable next-track glyph. One read of Music.app every five seconds, shared by every open session; the clock ticks in between.
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ./register.tsx env writes: nothing; ./register.tsx env reads: TMPDIR
/today: the day's agenda for Claude and for you. Calendar events, E3 deadlines and Reminders are read through the macos and nycu plugin scripts, shown as one line above the prompt with a countdown to the next event, and handed to Claude as a context block so it knows your day.
root: CLAUDE.md at the plugin root is not loaded as project context. To ship context with your plugin, use a skill (skills/<name>/SKILL.md) instead.; ./register.tsx env writes: nothing; ./register.tsx env reads: CLAUDE_CONFIG_DIR, HOME, TMPDIR
Mods are Claude Code plugins that run JavaScript or TypeScript hooks inside your session. They can add a dashboard, open a pane, change tool behavior, or bring context into the conversation.
Mods are on by default in Claude Code 2.1.287 and later. Choose a mod, open its repository, and follow the author’s setup instructions. The API can change between releases.
Each entry records the hooks and API calls reported by claude plugin validate. Access levels describe the widest reach of those calls. They are not safety ratings.
Validation is a static check, not a runtime compatibility test. A passing result does not prove that a mod works or is safe. Check the source and its setup instructions before installing.
This scan used Claude Code 2.1.287 on Oct 2, 2026. Read the method.
AGENTS.md as project instructions, by one option. Under claude-md-or-agents-md a project with no instruction files of its own gets its AGENTS.md files instead: a prompt.context hook asks $.fs.ancestors for every AGENTS.md and .claude/AGENTS.md from the filesystem root down to the working directory and hands them to the engine as project instruction files, which it renders, announces and withholds from agents exactly as it does CLAUDE.md, skipping any file the engine already loaded by import or link; a tool.call hook on Read attaches the ones under the project root as the engine attaches a nested CLAUDE.md. Under claude-md-and-agents-md every AGENTS.md is loaded beside CLAUDE.md. Under managed-only the project's and the person's instruction files are dropped and the organization's managed ones kept. Under claude-md the engine's walk stands alone. claude-md-or-agents-md is the default.
Access & validation details
Observed events
the system prompt, subagent spawns, Read calls
Hooks
session.start, prompt.context {instructionFiles has {kind=?}, prompt.context, agent.spawn {fork=true}, tool.call {tool=Read}
The diff pane as a plugin: /diff opens the session's uncommitted changes beside the transcript, file by file with their hunks, refreshed as Claude edits and runs commands; the first edit opens it where the terminal is wide enough, and a file's ask button rides its hunks on the next prompt.
Security default for organizations: seated outermost, it keeps the organization's classic hooks, prompt content, settings and tool policy out of reach of the plugins a person installs, and adds no policy of its own; one managed option, allowManagedModsOnly, limits mods to the organization's.
Access & validation details
Observed events
classic hooks, the system prompt, every skill, subagent spawns
Plugin analytics: hooks $.telemetry's two events, so a plugin logs an event or marks a feature's use as a first-party row, sent in batches with the session's own credential.
types ./types/index.d.ts declares on $: $.telemetry; modules../register.ts: its hooks stand on the telemetry stream "anthropic" (a matcher names it, or a telemetry hook names no "to" and so stands on every stream), which is for the plugins built into the CLI; name the collector on each telemetry hook: on("telemetry.log", { to: "collector" }, hook); ./register.ts env writes: nothing; ./register.ts env reads: ALACRITTY_LOG, APP_URL, AWS_EXECUTION_ENV, AWS_LAMBDA_FUNCTION_NAME, AZURE_FUNCTIONS_ENVIRONMENT, BUILDKITE, C9_PID, C9_USER, CF_PAGES, CI, CIRCLECI, CLAUBBIT, CLAUDE_AGENT_SDK_VERSION, CLAUDE_CODE_ACTION, CLAUDE_CODE_CONTAINER_ID, CLAUDE_CODE_CUSTOM_OAUTH_URL, CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC, CLAUDE_CODE_ENTRYPOINT, CLAUDE_CODE_HOST_PLATFORM, CLAUDE_CODE_PROVIDER_MANAGED_BY_HOST, CLAUDE_CODE_REMOTE, CLAUDE_CODE_REMOTE_ENVIRONMENT_TYPE, CLAUDE_CODE_REMOTE_SESSION_ID, CLAUDE_CODE_SESSION_ACCESS_TOKEN, CLAUDE_CODE_TAGS, CLAUDE_CODE_USE_ANTHROPIC_AWS, CLAUDE_CODE_USE_ANTHROPIC_GOOGLE_CLOUD, CLAUDE_CODE_USE_BEDROCK, CLAUDE_CODE_USE_FOUNDRY, CLAUDE_CODE_USE_GATEWAY, CLAUDE_CODE_USE_MANTLE, CLAUDE_CODE_USE_VERTEX, CLAUDE_CODE_WEBSOCKET_AUTH_FILE_DESCRIPTOR, CLAUDE_CONFIG_DIR, CLAUDE_SESSION_INGRESS_TOKEN_FILE, CODER, CODER_WORKSPACE_NAME, CODESPACES, COMSPEC, CURSOR_TRACE_ID, ConEmuANSI, ConEmuPID, ConEmuTask, DAYTONA_WS_ID, DENO_DEPLOYMENT_ID, DEVPOD, DEVPO… [+980 chars]; ./register.ts hooks telemetry for: anthropic, collector
network, runs processes, reads files, reads env vars, reads settings, reads the transcript, drawsfails on 2.1.287