collect-subagent-results-into-a-report
Use when a fan-out has finished and many child outputs must become one parent report. Aggregates by iterating the roster so totals are exact and per-row provenance survives.
- License 许可
- MIT
- 最近更新
- 2026-11-13
- Stars
- ★ 81
- Forks
- 17
01 前言 Front Matter
- name
- collect-subagent-results-into-a-report
- description
- Use when a fan-out has finished and many child outputs must become one parent report. Aggregates by iterating the roster so totals are exact and per-row provenance survives.
02 资源文件 3
- 许可证LICENSE1.0 KB
- 参考文档SOURCE.md364 B
- 参考文档TRUST.auto.yaml2.1 KB
03 SKILL.md 正文
Collect Subagent Results into a Report
The parent's report is a reduction over children, and the reduction must be mechanical. Iterate the roster in code, never assemble the summary from what you remember seeing.
Procedure
- Load the roster from
notes/roster.jsonand treat it as the authoritative list of expected results. - For every roster entry, append the child's verified artifact to
out/report/rows.jsonl, in roster order. - Assert coverage before summarising:
jq length notes/roster.jsonequals the row count; a shortfall is a missing child, not an empty one. - Aggregate with code:
jq -s 'group_by(.area) | map({area:.[0].area, n:length})' out/report/rows.jsonl. - Include only values you verified yourself (per
notes/metrics.tsv), not the child's raw claims. - State totals as counts you can reproduce, and attach the command behind each: "20/20 valid (validate.py)".
- Preserve provenance per row as
{slug, child_id, artifact_path, verdict}so any line traces to its child. - List failures explicitly with their blocking reason; do not bury them under the success summary.
- Keep the report in a fixed order the requester asked for; if none was asked, use roster order and say so.
- Re-read the final report against the roster once, as a last check, before delivering it.
Pitfalls
- Writing the summary from memory of the children's messages instead of iterating the roster file.
- Reporting the children's claimed totals rather than recomputed ones, propagating inflated numbers.
- Dropping a failed or missing child so the report reads cleaner than reality.
- Sorting or grouping by eye and miscounting, when
jqorsort | uniq -cgives the exact figure. - Losing provenance, so a wrong row in the report cannot be traced back to the child that produced it.
Verification
jq length notes/roster.json; wc -l < out/report/rows.jsonl
# passes when the two numbers match and every failure has an explicit reason line
Report to the user: the roster size, the reproduced aggregate, and any child that failed with its reason.
04 同仓库更多 Skills
recipe-performance
Diagnose a specific MetaMask user flow's real-app performance and Sentry critical-user-flow telemetry with deterministic mm-harness recipes. Use for CUF boundaries, cold/warm/background-resume behavior, latency regressions, lifecycle tags, duplicate or blocking network requests, long tasks, layout/paint cost, screen redraws, React rerenders, or browser trace analysis where tests alone cannot prove the real app.
mexc-spot-websocket
MEXC Spot WebSocket workflows for public streams, protobuf handling, depth maintenance, listen keys, and private streams without exposing API keys.
mexc-spot-rest
MEXC Spot REST workflows for public data, signed account requests, orders, wallet/admin actions, and secure signing without exposing API keys.
mexc-futures-websocket
MEXC Futures WebSocket workflows for public streams, private account/order/position streams, ping/pong, subscriptions, and depth maintenance.
0xklen-version-and-refresh-sanctions-lists
暂无简介
verify-tests-cover-changed-lines
Use when a PR adds code that the existing suite may not exercise. Measures patch coverage — the share of newly changed lines hit by tests — before approving.