Skill monitor

Skill wei63w/skill-monitor/skills/skill-monitor

Track Agent Skill usage, estimated tokens, Rules/AGENTS.md bloat, and Task/sub-agent startup cost. Install: npx skills add wei63w/skill-monitor

Install

npx -y skills add wei63w/skill-monitor --skill skill-monitor

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

17 days oldThe repository was created 17 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Monitor local Agent Skill usage, Rules/AGENTS.md context bloat, and Task / sub-agent startup cost (starts × system-prompt volume). Use for skill frequency, skill token estimates, unused skills, context bloat, AGENTS.md / rules size, subagent cost, backfill from transcripts, or project skill tracking. Also use for skill 监控, skill 统计, skill 频率, Rules 膨胀, AGENTS.md 体积, 子 Agent / Task 成本.

SKILL.md

2.6 KB, 586 tokens by cl100k_base, as published. Nobody here has run it

Skill Monitor

Track skill usage, always-on context bloat, and sub-agent startup cost in the project-local data/ directory. Capture is best-effort.

Quick setup

node .cursor/skills/skill-monitor/scripts/cli.mjs setup --project . --hooks
node .cursor/skills/skill-monitor/scripts/cli.mjs backfill
node .cursor/skills/skill-monitor/scripts/cli.mjs backfill-tasks
node .cursor/skills/skill-monitor/scripts/cli.mjs analyze --project . --write
node .cursor/skills/skill-monitor/scripts/cli.mjs bloat --project . --snapshot --write
node .cursor/skills/skill-monitor/scripts/cli.mjs analyze-tasks --project . --write

setup --hooks installs:

beforeReadFile → record SKILL.md loads
subagentStart + preToolUse (Task) → record sub-agent starts with Rules/AGENTS volume

Commands

Command	Purpose
`setup [--hooks]`	Init `data/`; merge Cursor hooks
`record --path <SKILL.md>`	Record one skill use
`analyze [--write]`	Skill frequency + est. tokens
`bloat [--snapshot] [--write]`	Rules / AGENTS.md bloat + budgets
`record-task --type <name>`	Record one Task / sub-agent start
`analyze-tasks [--write]`	Startup cost: starts × avg prompt volume
`backfill`	Skills from Cursor transcripts
`backfill-tasks`	Task/subagent starts from transcripts
`reestimate`	Refresh skill `estTokens`
`list-skills` / `summary`	Discovery / aggregates

Data layout

data/events.jsonl — skill loads (estTokens)
data/summary.json — skill aggregates
data/tasks.jsonl — sub-agent starts (systemPromptTokens)
data/context-snapshots.jsonl — optional bloat history
data/report.md / bloat-report.md / tasks-report.md

Agent workflow

Skill stats → analyze --write
Context inflation → bloat --snapshot --write
Sub-agent cost → analyze-tasks --write (after hooks or backfill-tasks)
Mention limits: estimates are not provider billing; hooks are strongest in Cursor.

Manual fallbacks

node .cursor/skills/skill-monitor/scripts/cli.mjs record --path "C:/path/to/skill/SKILL.md"
node .cursor/skills/skill-monitor/scripts/cli.mjs record-task --type explore --project .

Gives 0 of the 12 instructions most prompt engineering skills give in 586 tokens

Counted across 563 of the 626 authors here whose files we hold, read 2026-08-06

ask at most three clarifying questionsin 22 of 563, across 15 files
respond in the user input languagein 14 of 563, across 9 files
preserve the original intentin 13 of 563, across 11 files
Establish baseline metrics and collect representative examplesin 12 of 563, across 2 files
Identify failure modes and prioritize high-impact fixesin 12 of 563, across 2 files
Apply prompt and workflow improvements with measurable goalsin 12 of 563, across 2 files
Roll back quickly if quality or safety metrics regressin 12 of 563, across 2 files
validate changes with tests and roll out in controlled stagesin 12 of 563, across 2 files
generate quantitative baseline performance reportsin 12 of 563, across 2 files
create representative test scenariosin 12 of 563, across 2 files
treat prompts as codein 12 of 563, across 5 files
test prompts on diverse inputsin 12 of 563, across 8 files

Said here and by no other author read

setup data directory and merge hooks
backfill skill loads from transcripts
analyze skill frequency and write output
snapshot rules bloat and write output
analyze sub-agent startup cost and write output
mention limits of token estimates and hooks

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.