Kimchi Validate

<command_purpose> Ensure each bead is standalone-executable by running 4 validators. Auto-enrich failing beads. Flag for human review if enrichment can't fix them. </command_purpose>

Input

Read all .beads/*.yaml files (excluding manifest.yaml).

Parse $ARGUMENTS:

• •--loops N: Max enrichment iterations per bead (default: 3)
• •--bead ID: Validate only one specific bead

Process

1. Run All Validators Per Bead

For each bead, run these 4 checks:

Context Completeness:

• •Every file reference has a find: strategy
• •Find terms are specific enough to be unique
• •Find terms are stable (class/method names, not comments or TODOs)
• •Scope is appropriate for purpose
• •Error handling patterns referenced when task involves error handling
• •FAIL if: any line number reference, implicit references, find terms matching multiple locations

Deliverables Clarity:

• •Each deliverable has explicit type (file_create, file_modify, file_delete)
• •file_create has full path
• •file_modify has find: anchor point
• •Descriptions are concrete (not "update the service")
• •FAIL if: missing path, file_modify without anchor, vague descriptions

Test Specification:

• •Test file path specified
• •Test cases are scenarios, not vague ("uploads file" not "test uploading")
• •Edge cases included (minimum by complexity: S=2, M=4, L=6)
• •Run command provided
• •FAIL if: no test file, fewer than minimum cases, no run command

Isolation Check:

• •No references to other beads' internal details
• •Dependencies only reference bead IDs
• •All context is embedded or findable via find: landmarks
• •No pronouns without antecedents ("it", "the others")
• •Acceptance criteria don't reference other beads' outputs
• •FAIL if: prose references to other beads, missing context

2. Enrichment Loop

For each failing bead:

while (score < passing) and (iterations < max_loops):
    identify what's missing
    enrich the bead:
      - Add missing context references (search codebase for find: terms)
      - Add missing find: anchors for file_modify deliverables
      - Add test cases to meet minimum
      - Replace vague descriptions with concrete ones
      - Resolve cross-bead references to actual file references
    re-validate

3. Flag Unrepairable Beads

If a bead can't pass after max_loops:

• •Flag it in the report
• •List specific failures
• •Suggest manual fixes

4. Write Validation Report

Write .kimchi/VALIDATION-REPORT.md:

# Bead Validation Report

**Validated:** [today's date]

## Summary

| Bead | Context | Deliverables | Tests | Isolation | Status |
|------|---------|-------------|-------|-----------|--------|
| 001 | PASS | PASS | PASS | PASS | PASS |
| 002 | ENRICHED | PASS | PASS | PASS | PASS |
| 003 | PASS | PASS | ENRICHED | PASS | PASS |

## Enrichments Applied

### Bead [ID]
- [What was enriched and why]

## Failed Beads (Require Manual Fix)

### Bead [ID]
- [What failed and suggested fix]

## Result

[N] of [N] beads validated and ready.

5. Offer Push

If all beads pass, ask: "Push beads to beads-sync branch? [y/n]"

If yes:

git add .beads/
git commit -m "kimchi: validated beads for [feature name]"
git push origin HEAD:beads-sync

Report: "[N] beads validated. Saved report to .kimchi/VALIDATION-REPORT.md"