telemetry-review
Skill0 starsChanged 3 days agoMITView on GitHub
Review 2build's own telemetry, not product analytics. Use to turn ~/.babysit/analytics (skill-usage.jsonl, decisions.jsonl) into a short ticket-ready report — which skills fire and fail, whether plan-draft habitually over-sizes, where runs go BLOCKED.
In this repository30 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- llms.txtllms.txt
- agent-first-docsSkill
- analytics-reviewSkill
- architectSkill
- autopilotSkill
- browseSkill
- create-prSkill
- design-uiSkill
- foremanSkill
- harness-auditSkill
- humanizer-viSkill
- image-genSkill
- implementSkill
- investigateSkill
- marketing-videoSkill
- perf-auditSkill
- plan-draftSkill
- qaSkill
- reconSkill
- release-prdSkill
- releaseSkill
- review-prSkill
- semantic-decisionSkill
- setup-projectSkill
- teamSkill
- testSkill
- token-auditSkill
- triageSkill
- ui-auditSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Default window is the last 30 days.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
telemetry-review
A maintainer pass over babysit itself. Telemetry is the pack's primary feedback channel; this skill closes the loop by reading it and emitting tickets, not dashboards. Read-only — it never edits the pack in the same run.
Flow
- Locate the data:
A="${BABYSIT_ANALYTICS_DIR:-$HOME/.babysit/analytics}". Default window is the last 30 days; honor an explicit window from the invocation. Empty/missing files are a valid finding ("telemetry not flowing"), not an error. - Aggregate
skill-usage.jsonl: runs / error rate / median duration per skill (event:"end"rows carryoutcome+duration_s), plus the audit events (skill-verdict-audit, Hook C rows). Flag skills with error rate ≥25% with at least five completed runs. Compare against the installed skill inventory before reporting zero usage; low usage alone is not a defect. - Aggregate
decisions.jsonl.kind:"resize"rows → isplan-drafthabitually over-sizing? - Find where runs die:
outcome:"error"clusters, BLOCKED/NEEDS_CONTEXT in verdict audits, and tickets whose checkpoint stopped advancing (if ticket state is reachable). Pairstart/endrows byinvocation_idwhen present (the current runtime schema); legacy rows fall back to(session, skill). Find unmatched starts, not presumed crashes. Check live session/ticket state and window boundaries before calling a run orphaned; separate unknown completion from observed failure. Pair repeated starts/ends chronologically rather than collapsing all invocations into one pair. Pair within the window only: session ids drifted in older history, so cross-history grossstarts − endsis unreliable (it can even go negative at a window edge). A skill that emitsstartbut never anyendmay indicate an instrumentation gap — report that uncertainty separately from confirmed run failures. - Emit the report: a ranked, ticket-ready list. Each item is one line of ticket title + one line of evidence (counts, skill names, sample ts). ≤7 items; below that bar, say "no action needed" — an empty report is a valid result.
Rules
- Evidence or it didn't happen: every item cites counts from the files, never impressions. Don't invent trends from <5 data points.
- Report, don't fix. Changing a skill because of a finding is the next ticket
(usually
implement), not this run. - Local-only data: never send telemetry contents to an external service.
Output
STATUS: DONE | NEEDS_CONTEXT | BLOCKED
VERDICT: AUDITED
REPORT:
1. <ticket-ready title> — <evidence: counts, skills, window>
...
NEXT: file the top item as a ticket, or "no action needed"Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

