菜单
Manage and report the opt-in, privacy-clean usage telemetry beacon. Use when the user asks to "enable/disable telemetry", "show telemetry", "usage stats", "which commands do I use", or "how often is the commit gate bypassed".
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/bdfinst/agentic-dev-team --skill telemetry命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Validate test suite quality by running a real mutation testing tool and triaging surviving mutants. Use after writing tests to verify assertions catch behavioral changes, when evaluating test coverage quality, or as a CI quality gate on critical modules. The AI value here is triage — classifying survivors, writing fix tests — not generating or estimating mutations.
Execute an approved implementation plan in small per-behavior batches. Reads the plan, implements each step one behavior at a time in the Code-First Small Batches cadence with a refactor on every green, runs inline review checkpoints, and produces verification evidence. Use when the user says "build this", "implement the plan", "start building", or after /plan has been approved.
Run all enabled review agents against target files. Use this whenever the user asks for a code review, wants feedback on their code, says "review my code", "check this before I PR", "what's wrong with this", "run the agents", or has just finished implementing a feature. Use proactively before commits and pull requests.
基于 SOC 职业分类
正在显示 SKILL.md
| name | telemetry |
| description | Manage and report the opt-in, privacy-clean usage telemetry beacon. Use when the user asks to "enable/disable telemetry", "show telemetry", "usage stats", "which commands do I use", or "how often is the commit gate bypassed". |
| argument-hint | [on|off|status|report] |
| user-invocable | true |
| allowed-tools | Bash(python3 *, jq *, cat *, mkdir *, ls *), Read, Write |
Role: worker. Manages consent for and reports the local, opt-in telemetry
beacon. The beacon (hooks/telemetry.py) records MINIMAL events — a command
NAME, a skill NAME (including agent-/auto-invoked skills), a gate name +
outcome, and the plugin version — to ~/.claude/metrics/telemetry.jsonl. No
prompts, paths, code, or payloads are ever recorded, and there is no network
egress: everything stays local.
Telemetry is OFF by default. It activates only when the user-level
~/.claude/telemetry.json contains {"enabled": true}. This is a
config-file-only, home-scoped switch: it is read and written at this one path
regardless of which project directory the command runs from.
This beacon is intentionally a cheap, always-on local counter: which
commands/skills run, how often the commit gate is bypassed. It is not the
self-improvement loop and must not grow into one. The loop is /session-review,
whose digest already extracts a superset of these signals (token, rework,
accuracy, utilization) from transcripts and routes suggestions to governed
machinery. Cross-machine aggregation also belongs to the session-digest
(Delta D / #178), not this beacon. If you find yourself adding analysis or
network egress here, that work belongs in /session-review instead.
status (default): report whether telemetry is enabled and what's collected.on: enable by writing ~/.claude/telemetry.json with {"enabled": true}
(confirm with the user first — this starts local recording).off: disable by writing {"enabled": false} (recording stops; the existing
log is left in place for the user to inspect or delete).report: print the usage summary.status — check ~/.claude/telemetry.json; state on/off, and list
exactly what is and isn't collected (events: command/skill name, gate
fired/bypassed, plugin version; never: prompt text, paths, code, network).
on / off — write ~/.claude/telemetry.json:
{ "enabled": true }
For on, confirm consent first. Recording is local-only; the event log
~/.claude/metrics/telemetry.jsonl lives outside any project directory so
it can never be committed.
report — summarize the event log:
python3 "${CLAUDE_PLUGIN_ROOT}/hooks/lib/telemetry_report.py" \
--log "$HOME/.claude/metrics/telemetry.jsonl"
Shows command usage, skill usage (including agent-/auto-invoked skills, counted distinctly from user-typed commands), and the pre-commit review gate's bypass rate. If the log doesn't exist, the report says telemetry is off and nothing has left the machine.
Report exactly what the tool emits; do not invent counts.