mcpField
S
skill-trace-debugger

skill-trace-debugger

github.com/wyattowalshTesting & quality535 installsUpdated 8w ago
Repository

About skill-trace-debugger

A repository skill that checks static trace and evaluation readiness for skills before live behavioral runs. It reports coverage signals for one skill or the full repository without executing evaluations or collecting session logs.

Key features

  1. 01Check trace readiness for one skill
  2. 02Summarize readiness across the repository
  3. 03Check evaluation and validator signals
  4. 04Identify explicit-invocation coverage gaps
  5. 05Provide machine-readable reports

Use cases

  1. 01Review a skill before behavioral evaluation
  2. 02Find trace coverage gaps across skills
  3. 03Check for missing readiness signals
  4. 04Generate a repository-wide readiness report

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/wyattowalsh/agents --skill skill-trace-debugger

After installing, type/skill-trace-debuggerin Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

evals/
evals.json
scripts/
check.py
trace_report.py
scripts/asset_toolkit/
validate_hooks.py

Details

From the project README

Skill Trace Debugger

Surface trace and eval coverage gaps for repository skills before live behavioral runs.

Scope: Static trace readiness checks only. Does not execute live evals or harvest session logs.

Dispatch

$ARGUMENTSAction
<name>Trace readiness report for one skill
repo / --allFleet trace readiness summary
--format jsonMachine-readable report
EmptyShow trace signal checklist

Trace Readiness Checklist

SignalMeaning
evals/evals.jsonBehavioral cases exist for post-hoc grading
scripts/check.pyPortable validator contract is wired
explicit-invocation caseEval includes a slash-command or named dispatch prompt
NOT-for boundaryDescription names the correct alternate skill

Workflow

uv run python skills/skill-trace-debugger/scripts/trace_report.py <name>
uv run python skills/skill-trace-debugger/scripts/trace_report.py --all --format json

Route live trace grading to agent-eval-runner or maintainer CI workflows.

Validation Contract

uv run python skills/skill-trace-debugger/scripts/check.py

Critical Rules

  1. Never claim live trace capture from static reports alone.
  2. Treat missing explicit-invocation eval cases as a coverage gap, not a hard failure for internal skills.
  3. Do not mutate eval manifests without maintainer approval.
  4. Pair trace reports with skill-eval-scaffolder when adding new cases.
  5. Keep prompts grounded in each skill's dispatch table.

Comments

Markdown supported