test-log-centralizer

星标0

分支0

更新时间2026年5月5日 23:52

Centralize test-run logs under <project>/.test-runs/<UTC-timestamp>/ so every other testing skill (auto-test, unit-test-runner, browser-test, integration-test-runner, flaky-detector, coverage-reporter) writes into a single shared run folder. Use ONLY as a dependency invoked by another testing skill — it is not user-facing on its own. Trigger when another skill needs to (a) start a new run folder via `scripts/init-run.sh`, (b) append per-layer log lines via `scripts/append-log.sh`, or (c) finalize a run via `scripts/finalize-run.sh`. Do NOT trigger on user phrases like "show me the logs", "tail the logs", "check what tests printed" — those should call `auto-test` (which internally uses this skill) or read `<project>/.test-runs/latest/run.log` directly with the Read tool. Do NOT use this skill for non-test logging (application logs, build output, lint output) — those belong to other tools.

安装

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。

来源

hieutrtr

hieutrtr/auto-test-skills

打开 GitHub 仓库查看创作者相关仓库

下载

test-log-centralizer

Singleton dependency for all testing skills in this repo. Provides a uniform path convention and three small bash entrypoints under scripts/ so every test layer (unit, integration, browser, orchestrator, skill, setup, teardown) writes into the same <project>/.test-runs/<run-id>/ folder. Output structure is documented in docs/ARCHITECTURE.md §(b).

When to use

Activate only when invoked as a sub-step of another testing skill — typically auto-test, unit-test-runner, browser-test, integration-test-runner, flaky-detector, or coverage-reporter. Concretely:

A caller skill needs to start a fresh run → run scripts/init-run.sh.
A caller skill needs to append stdout/stderr to a per-layer stream → run scripts/append-log.sh. Each emitted line is prefixed with an ISO-8601 UTC timestamp (ms-precision when python3 is available, second-precision otherwise).
A caller skill needs to finalize the run (merge streams → run.log, emit summary.json (final form) + manifest.json + run.json, re-point latest) → run scripts/finalize-run.sh.

Do not activate when the user message is about reading or tailing existing logs — that is a plain Read / Grep of <project>/.test-runs/latest/run.log and does not need this skill.

Do not activate for non-test logging (general application logs, build output, lint output, deployment output). Those belong elsewhere.

How to use

SKILL_DIR="$(dirname "$0")/.."   # caller adjusts to its own layout

# 1. Start a new run — echoes the absolute run-folder path on stdout.
RUN=$("$SKILL_DIR/scripts/init-run.sh" "<project-root>" "<runner>" "<command>")
# RUN now points to <project-root>/.test-runs/<UTC-ts>/.
# Initial folder contents (T-1.1):
#   run.log          (empty file; finalize-run.sh fills via stream merge)
#   meta.json        (run_id, start_ts, project_dir, runner, command, schema_version)
#   summary.json     (placeholder; finalize-run.sh promotes to final form)
#   streams/         (per-layer log files; written by append-log.sh)
#   screenshots/     (browser-test artifacts — Phase 2)

# 2. Append per-layer log lines.
# Inline form:
"$SKILL_DIR/scripts/append-log.sh" "$RUN" unit "starting suite tests/auth"
# Stdin form (stream a runner's stdout/stderr line-buffered):
bun test 2>&1 | "$SKILL_DIR/scripts/append-log.sh" "$RUN" unit
# Each appended line is also tee-ed to streams/all.log for live tail.

# 3. Finalize — merge streams into run.log, write final summary/manifest/run.json,
# re-point <project>/.test-runs/latest. Counts are caller-supplied (T-1.4 parsers
# fill them later); pass 0/0/0/0 for a placeholder.
"$SKILL_DIR/scripts/finalize-run.sh" "$RUN" "$EXIT_CODE" "$TOTAL" "$PASSED" "$FAILED" "$SKIPPED"

# 4. (Optional) Trigger retention sweep at end-of-run by exporting TEST_LOG_RETAIN=1
# before finalize. Default OFF so test suites for T-1.1..T-1.5 stay isolated; the
# auto-test orchestrator opts in. Equivalent to invoking retain.sh directly:
"$SKILL_DIR/scripts/retain.sh" "<project-root>/.test-runs"
# Policy: keep newest 10 plain, gzip the next 2 (run.log → run.log.gz, drop
# streams/ + screenshots/, keep run.json + manifest.json + summary.json + meta.json),
# delete anything older. Idempotent on re-run.

The <runner> and <command> arguments to init-run.sh are optional — when omitted, meta.runner defaults to "tbd" and meta.command to the empty string. They are filled later by the caller skill once framework detection has run. See references/schema.md for the full JSON shape contract that downstream skills + agents can rely on.

Examples

User says: "run the unit tests" → Claude invokes auto-test, which calls unit-test-runner, which calls this skill's init-run.sh to scaffold the run folder before spawning the framework.

User says: "show me the latest test log" → Claude does not invoke this skill; it calls Read on <project>/.test-runs/latest/run.log directly.

Files

scripts/
  init-run.sh           — T-1.1: scaffold run folder + meta.json + summary.json placeholder
  append-log.sh         — T-1.2: per-layer stream append with ISO-8601 UTC ts prefix
  finalize-run.sh       — T-1.2: merge streams → run.log, write summary/manifest/run.json,
                          re-point .test-runs/latest atomically; optional retain hook (T-1.6)
  retain.sh             — T-1.6: count-based retention sweep (10 plain + 2 gz + prune)
references/
  schema.md             — T-1.2: contract for meta.json, summary.json, manifest.json, run.json
tests/
  test-init-run.sh      — T-1.1 acceptance (35 assertions)
  test-append-log.sh    — T-1.2 acceptance (36 assertions)
  test-finalize-run.sh  — T-1.2 acceptance (55 assertions, incl. golden compare)
  test-retention.sh     — T-1.6 acceptance (71 assertions, incl. 12-/14-/5-run scenarios + idempotency + finalize wiring)
  goldens/
    manifest.golden.json — normalized golden for finalize-run shape stability
  run-all.sh            — convenience runner; bash run-all.sh exits 0 when everything green

Run a project's automated tests end-to-end and return a single structured report (ASCII dashboard + machine-readable trailing lines + path to .test-runs/<UTC-ts>/run.log). Trigger when the user asks to run, execute, or check the status of a project's tests — phrases like "run the unit tests", "run the tests", "execute the test suite", "ai check tests for me", "run npm test and tell me what failed", or "how are the tests doing?". This is the meta-skill that delegates to layer-specific runners (Phase 1 ships only the `unit-test-runner` child); future phases will add `browser-test`, `integration-test-runner`, and `coverage-reporter` to the same orchestration. Do NOT trigger on "tail the log", "show me the latest test output", or "what does .test-runs/latest contain?" — those are plain `Read` of `<project>/.test-runs/latest/run.log`. Do NOT trigger on lint, typecheck, build, deploy, or coverage-only requests — those belong to other (separate) skills. Do NOT trigger on browser-only / Playwright-with-page.goto navi

2026-05-060

desc-too-short

hieutrtr/auto-test-skills

too short

2026-05-060

totally-different-name

hieutrtr/auto-test-skills

Folder is `name-mismatch/` but the frontmatter declares `name: totally-different-name`. Linter check 6 must catch this. Long enough to clear the 40-char gate.

2026-05-060

no-body

hieutrtr/auto-test-skills

Frontmatter is fine and the description is well over 40 characters, but the body below the closing `

2026-05-060

unit-test-runner

hieutrtr/auto-test-skills

Detect the JavaScript / TypeScript unit-test framework in a project (Jest, Vitest, bun:test, Mocha, or @playwright/test used as a unit-test runner) and prepare a canonical run command. Trigger when another testing skill — usually `auto-test` — needs to know "what unit-test framework does this repo use, what command should I run, and which JS runtime / package manager (npm, pnpm, yarn, bun) drives it?". Concretely fires on user phrases like "run the unit tests", "what framework do these tests use?", "detect the test runner for this project", or as a sub-step of `auto-test`. Do NOT trigger on "run end-to-end tests" / "browser tests" / "integration tests" / "Playwright e2e tests with a dev server" — those belong to the Phase 2 `browser-test` skill (this skill's `playwright-runner` framework keyword is reserved for `@playwright/test` used WITHOUT browser navigation, i.e. as a generic test runner). Do NOT trigger on Python (pytest), Rust (cargo test), Go (`go test`), Ruby (rspec), or Elixir tests — those are Phase

2026-05-050

hello-test

hieutrtr/auto-test-skills

Phase-0 spike probe used to verify Claude Code skill discovery and triggering. Trigger ONLY when the user message contains the literal phrase "hello-test ping" — no paraphrases, no partial matches. Output is a single fixed acknowledgement string with the current ISO-8601 UTC timestamp, produced by running a one-liner Bash command. Do NOT trigger on the words "hello", "test", "ping" individually, on phrases like "hello world", "say hello", "ping me", or any other variant. This skill exists purely as a fingerprint for the auto-test-skills repo Phase 0 spike to confirm that user-level skills dropped at ~/.claude/skills/<name>/SKILL.md are auto-discovered. Safe to delete after Phase 0 sign-off.

2026-05-050

name	test-log-centralizer
description	Centralize test-run logs under <project>/.test-runs/<UTC-timestamp>/ so every other testing skill (auto-test, unit-test-runner, browser-test, integration-test-runner, flaky-detector, coverage-reporter) writes into a single shared run folder. Use ONLY as a dependency invoked by another testing skill — it is not user-facing on its own. Trigger when another skill needs to (a) start a new run folder via `scripts/init-run.sh`, (b) append per-layer log lines via `scripts/append-log.sh`, or (c) finalize a run via `scripts/finalize-run.sh`. Do NOT trigger on user phrases like "show me the logs", "tail the logs", "check what tests printed" — those should call `auto-test` (which internally uses this skill) or read `<project>/.test-runs/latest/run.log` directly with the Read tool. Do NOT use this skill for non-test logging (application logs, build output, lint output) — those belong to other tools.
allowed-tools	Bash, Read

test-log-centralizer

When to use

A caller skill needs to start a fresh run → run scripts/init-run.sh.
A caller skill needs to append stdout/stderr to a per-layer stream → run scripts/append-log.sh. Each emitted line is prefixed with an ISO-8601 UTC timestamp (ms-precision when python3 is available, second-precision otherwise).
A caller skill needs to finalize the run (merge streams → run.log, emit summary.json (final form) + manifest.json + run.json, re-point latest) → run scripts/finalize-run.sh.

Do not activate when the user message is about reading or tailing existing logs — that is a plain Read / Grep of <project>/.test-runs/latest/run.log and does not need this skill.

Do not activate for non-test logging (general application logs, build output, lint output, deployment output). Those belong elsewhere.

How to use

SKILL_DIR="$(dirname "$0")/.."   # caller adjusts to its own layout

# 1. Start a new run — echoes the absolute run-folder path on stdout.
RUN=$("$SKILL_DIR/scripts/init-run.sh" "<project-root>" "<runner>" "<command>")
# RUN now points to <project-root>/.test-runs/<UTC-ts>/.
# Initial folder contents (T-1.1):
#   run.log          (empty file; finalize-run.sh fills via stream merge)
#   meta.json        (run_id, start_ts, project_dir, runner, command, schema_version)
#   summary.json     (placeholder; finalize-run.sh promotes to final form)
#   streams/         (per-layer log files; written by append-log.sh)
#   screenshots/     (browser-test artifacts — Phase 2)

# 2. Append per-layer log lines.
# Inline form:
"$SKILL_DIR/scripts/append-log.sh" "$RUN" unit "starting suite tests/auth"
# Stdin form (stream a runner's stdout/stderr line-buffered):
bun test 2>&1 | "$SKILL_DIR/scripts/append-log.sh" "$RUN" unit
# Each appended line is also tee-ed to streams/all.log for live tail.

# 3. Finalize — merge streams into run.log, write final summary/manifest/run.json,
# re-point <project>/.test-runs/latest. Counts are caller-supplied (T-1.4 parsers
# fill them later); pass 0/0/0/0 for a placeholder.
"$SKILL_DIR/scripts/finalize-run.sh" "$RUN" "$EXIT_CODE" "$TOTAL" "$PASSED" "$FAILED" "$SKIPPED"

# 4. (Optional) Trigger retention sweep at end-of-run by exporting TEST_LOG_RETAIN=1
# before finalize. Default OFF so test suites for T-1.1..T-1.5 stay isolated; the
# auto-test orchestrator opts in. Equivalent to invoking retain.sh directly:
"$SKILL_DIR/scripts/retain.sh" "<project-root>/.test-runs"
# Policy: keep newest 10 plain, gzip the next 2 (run.log → run.log.gz, drop
# streams/ + screenshots/, keep run.json + manifest.json + summary.json + meta.json),
# delete anything older. Idempotent on re-run.

Examples

User says: "run the unit tests" → Claude invokes auto-test, which calls unit-test-runner, which calls this skill's init-run.sh to scaffold the run folder before spawning the framework.

User says: "show me the latest test log" → Claude does not invoke this skill; it calls Read on <project>/.test-runs/latest/run.log directly.

Files

scripts/
  init-run.sh           — T-1.1: scaffold run folder + meta.json + summary.json placeholder
  append-log.sh         — T-1.2: per-layer stream append with ISO-8601 UTC ts prefix
  finalize-run.sh       — T-1.2: merge streams → run.log, write summary/manifest/run.json,
                          re-point .test-runs/latest atomically; optional retain hook (T-1.6)
  retain.sh             — T-1.6: count-based retention sweep (10 plain + 2 gz + prune)
references/
  schema.md             — T-1.2: contract for meta.json, summary.json, manifest.json, run.json
tests/
  test-init-run.sh      — T-1.1 acceptance (35 assertions)
  test-append-log.sh    — T-1.2 acceptance (36 assertions)
  test-finalize-run.sh  — T-1.2 acceptance (55 assertions, incl. golden compare)
  test-retention.sh     — T-1.6 acceptance (71 assertions, incl. 12-/14-/5-run scenarios + idempotency + finalize wiring)
  goldens/
    manifest.golden.json — normalized golden for finalize-run shape stability
  run-all.sh            — convenience runner; bash run-all.sh exits 0 when everything green

test-log-centralizer

test-log-centralizer

When to use

How to use

Examples

Files

See also

test-log-centralizer

When to use

How to use

Examples

Files

See also

test-log-centralizer

test-log-centralizer

When to use

How to use

Examples

Files

See also

同仓库更多 Skills

同仓库更多 Skills

test-log-centralizer

When to use

How to use

Examples

Files

See also