Installs into .claude/skills of the current project.
Are you the author of Plan Validate?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/nhan03bit-plan-validate)
---
name: plan-validate
description: Validate implementation plans against intent using reverse-questioning, assumption stress-testing, and gap analysis
layer: hub
category: workflow
triggers:
- "/plan-validate"
- "validate this plan"
- "stress test the plan"
- "check the plan"
- "review the plan"
inputs:
- plan: The implementation plan to validate (from plan skill or user-provided)
- originalIntent: The original goal or request that spawned the plan (optional but recommended)
outputs:
- validationReport: Structured report with findings, gaps, and recommendations
- reverseQuestions: Questions that expose hidden assumptions or misalignments
- strengthScore: Qualitative assessment of plan readiness (Red/Yellow/Green)
linksTo:
- plan
- research
- scout
linkedFrom:
- plan
- cook
- team
- ship
preferredNextSkills:
- plan
- kanban
- scout
fallbackSkills:
- brainstorm
- research
riskLevel: low
memoryReadPolicy: selective
memoryWritePolicy: selective
sideEffects:
- Produces validation report
- May flag plan for revision
---
# Plan-Validate Skill
## Purpose
Serve as the adversarial reviewer of implementation plans. This skill's job is to find what is wrong, missing, or fragile in a plan BEFORE execution begins. It uses reverse-questioning, assumption stress-testing, and gap analysis to surface problems that are cheap to fix on paper and expensive to fix in code.
Think of this as a pre-mortem: imagine the plan has already failed, then work backward to find out why.
## Workflow
### Pass 1: Intent Alignment Check
1. **Restate the original intent** in one sentence. If the original intent is not available, infer it from the plan's objective.
2. **Compare plan scope to intent** -- Does the plan actually deliver what was asked for? Common misalignments:
- Plan is broader than intent (scope creep baked in)
- Plan is narrower than intent (missing features)
- Plan solves an adjacent problem (misunderstood requirement)
3. **Generate 3-5 reverse questions** -- Questions phrased as "If we complete this plan exactly as written, will we have [X]?" where X tests different facets of the original intent.
### Pass 2: Assumption Stress Test
4. **Extract every assumption** -- Both explicitly stated and implicit ones buried in the plan.
5. **For each assumption, apply the inversion test**: "What happens if the opposite is true?"
- If the answer is "the plan still works" -- the assumption is non-critical.
- If the answer is "the plan breaks" -- the assumption is load-bearing and must be verified before execution.
6. **Flag unverified load-bearing assumptions** as blockers.
### Pass 3: Completeness Audit
7. **Check for missing phases** -- Is there a gap between phases where work would need to happen but is not accounted for?
8. **Check for missing tasks within phases** -- Are there implied subtasks not listed?
9. **Check for missing cross-cutting concerns**:
- Error handling / failure modes
- Testing strategy
- Rollback plan
- Data migration (if applicable)
- Security implications
- Performance implications
- Accessibility (if UI work)
- Documentation updates
10. **Check dependency chain** -- Are there circular dependencies? Missing dependencies? Implicit ordering that is not stated?
### Pass 4: Feasibility Check
11. **Complexity calibration** -- Are the complexity estimates realistic? Look for:
- "Low complexity" phases with high uncertainty (likely underestimated)
- Phases with no unknowns (likely overconfident)
- Integration phases marked as simple (integration is rarely simple)
12. **Resource check** -- Does the plan assume resources (APIs, services, permissions) that may not be available?
13. **Timeline sanity** -- If a timeline is given, does it account for testing, review, and iteration?
### Pass 5: Verdict
14. **Assign a readiness score**:
- **Green**: Plan is solid. Minor suggestions only. Proceed to execution.
- **Yellow**: Plan has gaps or unverified assumptions that should be addressed first. Revise specific sections.
- **Red**: Plan has fundamental issues. Major revision needed. Return to `plan` skill.
15. **Produce the validation report** using the template below.
## Validation Report Template
```markdown
# Plan Validation Report
## Intent Alignment
**Original intent**: [one sentence]
**Plan delivers**: [one sentence summary of what the plan actually produces]
**Alignment**: Aligned | Partially Aligned | Misaligned
### Reverse Questions
1. If we complete this plan exactly, will [X]? → [Yes/No/Unclear]
2. ...
---
## Assumption Audit
| # | Assumption | Load-Bearing? | Verified? | Risk if Wrong |
|---|-----------|---------------|-----------|---------------|
| A1 | ... | Yes | No | [consequence] |
### Blockers (Unverified Load-Bearing Assumptions)
- A1: [assumption] — Must verify before Phase [N]
---
## Completeness Gaps
- [ ] [Missing element] — Affects Phase [N]
- [ ] ...
## Feasibility Concerns
- [concern 1]
- [concern 2]
---
## Verdict: 🟢 Green | 🟡 Yellow | 🔴 Red
### Recommendations
1. [Highest priority fix]
2. [Second priority fix]
3. ...
### Suggested Next Steps
- [ ] [action]
```
## Usage
### After creating a plan
```
/plan-validate
```
(Automatically picks up the most recent plan from context)
### With explicit plan reference
```
/plan-validate Check the authentication migration plan against the original requirement: "Users must be able to log in with Google and GitHub OAuth"
```
### Quick validation
```
/plan-validate Is this plan ready to execute?
```
## Examples
### Example: Catching scope misalignment
**Plan objective**: "Build user authentication with JWT"
**Original intent**: "Users need to log in"
**Reverse question**: "If we complete this plan, will users be able to reset their password?"
**Finding**: Plan covers login/signup but not password reset, which is implied by "users need to log in."
### Example: Catching a hidden assumption
**Plan task**: "Migrate user table to new schema"
**Hidden assumption**: Existing data can be transformed without loss
**Inversion**: What if some existing data does not fit the new schema?
**Finding**: Plan needs a data audit task before migration phase.
## Guidelines
- **Be constructively adversarial** -- The goal is to make the plan better, not to prove it is bad.
- **Prioritize findings** -- Not all gaps are equal. Flag blockers separately from nice-to-haves.
- **Do not rewrite the plan** -- Identify problems and let the user (or `plan` skill) fix them.
- **Ask the uncomfortable questions** -- "What happens when this fails?" is always a valid question.
- **One round is usually enough** -- If a plan fails validation badly, it needs replanning, not more validation passes.
- **Verify, do not assume** -- If you are unsure whether an assumption holds, flag it rather than guessing.