verification-plan-codex
Skill27 starsChanged 3 months agoMITView on GitHub
Use when a plan needs proof criteria before execution, especially coding changes, business deliverables, mixed business-coding projects, acceptance criteria, rollback triggers, manual checks, or final validation design.
In this repository16 other files it gives its agents.
- adversarial-plan-review-codexSkill
- canvas-design-codexSkill
- code-intelligenceSkill
- code-mapper-skillSkill
- deep-planning-codexSkill
- document-handoffSkill
- docx-winSkill
- frontend-design-codexSkill
- grill-me-codexSkill
- grill-with-docs-codexSkill
- handoff-codexSkill
- pptx-winSkill
- repo-map-codexSkill
- ultraplan-codexSkill
- web-artifacts-builder-codexSkill
- xlsx-winSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 14
Read success criteria, failure criteria, out-of-scope boundaries, repo map, evidence catalog, assumption ledger, and the current …
- Distinctive: line 23
Use the current plan, success criteria, failure criteria, and inspected evidence directly.
- Very specific: line 24
Use this table shape when a table improves clarity: Criterion, Proof method, Command/check, Expected result, Evidence artifact, …
- Distinctive: line 26
If a human decision is still required, identify the exact unresolved decision without adding a duplicate execution-approval gate.
- Distinctive: line 27
Do not write files while the current collaboration mode prohibits mutations.
- Distinctive: line 52
Do not accept vague checks such as "test it manually", "review output", "run the app", or "stakeholder approves" without defining the …
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Codex Verification Plan
Overview
Define how the work will be proven correct before execution starts. A verification plan maps each success criterion to commands, checks, expected results, evidence, and recovery triggers.
Workflow
- Read success criteria, failure criteria, out-of-scope boundaries, repo map, evidence catalog, assumption ledger, and the current implementation plan.
- Classify validation as software, business, or mixed.
- For every success criterion, define at least one proof method.
- For every failure criterion, define the signal that would show it happened.
- Define rollback or recovery triggers before execution.
- Return the verification plan in conversation. Write an artifact only when the user explicitly requests one, and place it in the active session folder or another user-approved destination.
Native Plan Mode
- Use the current plan, success criteria, failure criteria, and inspected evidence directly; do not require a repository planning workfolder.
- Use this table shape when a table improves clarity:
Criterion,Proof method,Command/check,Expected result,Evidence artifact,Owner/Reviewer,Failure signal, andRecovery trigger. - Map every success criterion to at least one proof method and every failure criterion to a failure signal.
- If a human decision is still required, identify the exact unresolved decision without adding a duplicate execution-approval gate.
- Do not write files while the current collaboration mode prohibits mutations.
References
Read as needed:
references/software-validation.mdfor tests, builds, static checks, smoke checks, and command output.references/business-validation.mdfor stakeholder acceptance, source reconciliation, document checks, and calculation checks.references/rollback-and-recovery.mdfor Git and no-Git recovery strategies.
Minimum Output
Include:
- criterion identifier;
- proof method;
- command or manual check;
- expected result;
- evidence artifact;
- owner or reviewer if human review is required;
- failure signal;
- rollback or recovery trigger.
Red Flags
Do not accept vague checks such as "test it manually", "review output", "run the app", or "stakeholder approves" without defining the specific evidence that makes the check pass.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

