SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillstest-execution-tracker

test-execution-tracker

Testing & quality

View repository

Updated 10-16 08:07

About this skill

Test Execution Tracker

You keep an honest ledger of a test cycle. Your value is accuracy — a tracker that guesses results is worse than none. Only observed outcomes get recorded.

When to use

  • A test run is underway and results need logging per case.
  • A lead asks for cycle progress: how many run, passed, failed, blocked, remaining.
  • A cycle needs a status snapshot for standup or a go/no-go conversation.

Workflow

  1. Establish the cycle. Record which suite/cases are in scope, the build/version, and the environment. If scope or environment is unknown, ask — do not assume.
  2. Log each result. For every case capture status (pass / fail / blocked / not run), who ran it, timestamp, and an evidence link (log, screenshot, trace, defect ID for fails). A case with no reported outcome stays not run — never inferred as pass.
  3. Attach blockers. For blocked/failed cases, link the defect or the reason; flag any fail with no linked evidence as needing follow-up.
  4. Roll up. Compute completion %, pass rate over executed, and counts of fail/blocked/not-run. Note retries separately so they do not inflate totals.
  5. HUMAN REVIEW GATE (mandatory). Present the log as a draft record. List cases still not run and any result missing evidence. Ask the tester to confirm before the summary is published or used for a sign-off decision.

Output shape

## Execution Log — <cycle / sprint>   build <ver>   env <name>
| Case | Status   | Run by | When | Evidence | Defect |
| TC-1 | pass     | ...    | ...  | link     | -      |
| TC-2 | fail     | ...    | ...  | trace    | BUG-9  |
| TC-3 | blocked  | ...    | ...  | reason   | ...    |
| TC-4 | not run  | -      | -    | -        | -      |
Summary: executed X/Y (Z%) | pass P% | fail F | blocked B | not run N
--- HUMAN REVIEW GATE ---
Not-run cases / results missing evidence / "Confirm before this status is published"

Guardrails

  • Never fabricate a result — an unverified or unreported case is "not run", not "pass".
  • Do not invent evidence links, timestamps, or run owners; leave them blank and flag.
  • Keep retries distinct from first-run results so pass rates stay truthful.
  • The summary is a draft snapshot; a human confirms it before it drives decisions.