codex-completion-contract

Estrelas12

Forks3

Atualizado20 de julho de 2026 às 16:53

Defines every required outcome and its evidence before acting on long-running work, then audits each requirement as proved/weak/missing/ contradicted. Only run when explicitly asked for a completion audit.

Instalação

Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.

Fonte

ayeshakhalid192007-dev

ayeshakhalid192007-dev/LoopEngineering-CrashCourse

Abrir repositório GitHub Ver repositórios do creator

Download

Ocupações relacionadasSOC

Baseado na classificação ocupacional SOC

Desenvolvedores de softwareInformática e Matemática·SOC 15-1252

SKILL.md

readonly

name	codex-completion-contract
description	Defines every required outcome and its evidence before acting on long-running work, then audits each requirement as proved/weak/missing/ contradicted. Only run when explicitly asked for a completion audit.

codex-completion-contract procedure

The prompt carries intent; this skill carries the procedure — keep it a checklist, not an essay. Adapted from Forward Future Loop Library #28 (3goblack) to this library's L1-first, report-only shape.

Before starting: only run this for a named Goal, and only when explicitly asked for a completion audit — ask before creating Goal state at all. This is not for ordinary task planning.

Define the contract — before any action, list every required outcome and what evidence would prove it (a passing test, a verified deploy log, a re-read diff — not a claim).
Read — the target work's actual current state.
Audit — mark every requirement: proved (real, current evidence exists), weak (evidence is partial or stale), missing (no evidence yet), or contradicted (evidence says it didn't work).
Decide:
- All proved → the Goal is complete.
- Any not proved, and the work can continue → note what's needed next.
- A requirement can't be proved right now → blocked, not failed.
- No new evidence for 3 consecutive beats → stalled.
- The run limit or token cap is reached first → exhausted. This is never the same as complete — budget exhaustion is not success.
Write completion-audit.md: the requirement-to-evidence table, status (complete/blocked/stalled/exhausted), owner, and next action.
Record — update codex-completion-contract-state.md, append one line to loop-run-log.md, commit both.
Take NO other action — writes beyond the audit + spine are out of scope at L1.

Mais deste repositório

mesmo repositório

stale-safe-batch-release

ayeshakhalid192007-dev/LoopEngineering-CrashCourse

Reviews pending PRs, excludes stale/unfinished work, combines the rest into a release batch built from main. Read-only — reports the batch, never actually releases.

2026-07-2012

spec-dev-review

ayeshakhalid192007-dev/LoopEngineering-CrashCourse

Writes a scoped spec packet for one ticket, reviewed by a genuinely independent adversarial pass that also regression-checks prior findings. Read-only outside the draft — never writes the canonical spec.

2026-07-2012

loop-harness-verification

ayeshakhalid192007-dev/LoopEngineering-CrashCourse

General-purpose harness for scheduled repo work — stage in an isolated worktree, verify with a genuinely separate second session, ship only after a pass. Read-only — reports, never delivers.

2026-07-2012

clodex-adversarial-review

ayeshakhalid192007-dev/LoopEngineering-CrashCourse

Runs an adversarial review cycle against a PR (Claude implements/fixes, Codex or a subagent stand-in reviews), up to 5 iterations. Read-only on the real PR — drafts fixes, never pushes them.

2026-07-2012

ticket-to-pr-ready

ayeshakhalid192007-dev/LoopEngineering-CrashCourse

Turns a ticket or bug report into a review-ready patch. Read-only — reproduces, roots out the cause, drafts a fix, reports it. Never opens a PR, never folds in unrelated refactors.

2026-07-2012

repo-cleanup-loop

ayeshakhalid192007-dev/LoopEngineering-CrashCourse

Inspects branches, PRs, commits, and worktrees for what's stale vs. valuable. Read-only. Never deletes uncertain work or someone else's PR without confirmation.

2026-07-2012

name	codex-completion-contract
description	Defines every required outcome and its evidence before acting on long-running work, then audits each requirement as proved/weak/missing/ contradicted. Only run when explicitly asked for a completion audit.

codex-completion-contract procedure

Before starting: only run this for a named Goal, and only when explicitly asked for a completion audit — ask before creating Goal state at all. This is not for ordinary task planning.

Define the contract — before any action, list every required outcome and what evidence would prove it (a passing test, a verified deploy log, a re-read diff — not a claim).
Read — the target work's actual current state.
Audit — mark every requirement: proved (real, current evidence exists), weak (evidence is partial or stale), missing (no evidence yet), or contradicted (evidence says it didn't work).
Decide:
- All proved → the Goal is complete.
- Any not proved, and the work can continue → note what's needed next.
- A requirement can't be proved right now → blocked, not failed.
- No new evidence for 3 consecutive beats → stalled.
- The run limit or token cap is reached first → exhausted. This is never the same as complete — budget exhaustion is not success.
Write completion-audit.md: the requirement-to-evidence table, status (complete/blocked/stalled/exhausted), owner, and next action.
Record — update codex-completion-contract-state.md, append one line to loop-run-log.md, commit both.
Take NO other action — writes beyond the audit + spine are out of scope at L1.