output-contract-enforcer {Core} (v1.0)

"If the contract says 5 sections, I count 5 sections. Not 4. Not 6."

Purpose

Post-execution validator that checks every skill's output against its declared Inputs/Outputs table. Verifies format compliance (markdown structure, evidence tags, naming conventions) and completeness (all required sections present). [EXPLICIT]

When to use:

• After every skill execution (implicit validation)
• Before gate evaluation (pre-gate check)
• When /jm:verify is invoked

---

Core Principles (Immutable Laws)

1. Law of Contract: Every skill declares its outputs in the I/O table. The output MUST match. [EXPLICIT]
2. Law of Evidence: Analysis outputs MUST contain evidence tags. No tags = contract violation. [EXPLICIT]
3. Law of Naming: Files follow conventions: kebab-case for files, PascalCase for components, camelCase for functions. [EXPLICIT]

---

Core Process (Step-by-Step)

Phase 1: Contract Loading

1. Read the skill's SKILL.md Outputs table. [EXPLICIT]
2. Extract expected outputs: file type, name pattern, required sections. [EXPLICIT]

Phase 2: Validation

1. Check existence: Does the output file exist?
2. Check format: Does it match the declared type (Markdown, JSON, HTML, etc.)?
3. Check sections: For Markdown outputs, are all required sections present?
4. Check evidence tags: For analysis outputs, are [CODE]/[CONFIG]/[DOC]/[INFERENCE]/[ASSUMPTION] present?
5. Check naming: Does the file follow naming conventions (R-008)?

Phase 3: Verdict

1. Pass: Output matches contract → proceed. [EXPLICIT]
2. Fail: List specific violations → block until fixed. [EXPLICIT]

---

3. Inputs / Outputs

Inputs

Input	Type	Required	Description
Skill SKILL.md	File	Yes	The skill definition with I/O table
Generated output	File(s)	Yes	The output to validate

Outputs

Output	Type	Description
Validation report	Text	Pass/fail per check with specifics

---

Validation Gate (10x Checklist)

• Contract loaded from skill's I/O table
• All declared outputs exist
• Format matches declared type
• Evidence tags present in analysis outputs
• Naming conventions followed (R-008)

---

5. Self-Correction Triggers

[!WARNING] IF output has zero evidence tags but skill is in analysis category THEN FAIL. Evidence tags are mandatory for analysis.

[!WARNING] IF output file uses wrong naming convention THEN auto-suggest the correct name but don't rename without user approval.

Usage

Example invocations:

• "/output-contract-enforcer" — Run the full output contract enforcer workflow
• "output contract enforcer on this project" — Apply to current context

Assumptions & Limits

• Assumes access to project artifacts (code, docs, configs) [EXPLICIT]
• Requires English-language output unless otherwise specified [EXPLICIT]
• Does not replace domain expert judgment for final decisions [EXPLICIT]

Edge Cases

Scenario	Handling
Empty or minimal input	Request clarification before proceeding
Conflicting requirements	Flag conflicts explicitly, propose resolution
Out-of-scope request	Redirect to appropriate skill or escalate