sdlc-checkpoint

Save, load, validate, and check status of SDLC artifacts for a TriStar feature. Manages checkpoint state across pipeline phases to enable session continuity.

SKILL.md

GitHub

SDLC Checkpoint Skill

Prime Directive

"Never lose progress. Never repeat work."

This skill manages pipeline state persistence so that SDLC phases can be resumed across sessions without losing context or repeating completed work.

Arguments

  • --feature=<name> (required) - Feature name for checkpoint management
  • Operation is the first positional argument: save, load, status, validate

Usage

/sdlc-checkpoint save --feature=weather-activation
/sdlc-checkpoint load --feature=weather-activation
/sdlc-checkpoint status --feature=weather-activation
/sdlc-checkpoint validate --feature=weather-activation

Operations

save

Persist the current state of all artifacts for the feature.

  1. Scan docs/artifacts/<feature>/ for spec artifacts
  2. Scan .claude/checkpoints/<feature>/ for execution artifacts
  3. Record which artifacts exist and their last-modified timestamps
  4. Record the current pipeline stage based on which artifacts are present
  5. Write checkpoint state to .claude/checkpoints/<feature>/checkpoint.json

load

Restore context from a saved checkpoint.

  1. Read .claude/checkpoints/<feature>/checkpoint.json
  2. Verify all referenced artifacts still exist at their paths
  3. Display current pipeline stage and next expected action
  4. Display summary of each artifact (title, date, key metrics)
  5. Return the loaded state for the calling skill to use

status

Display the current pipeline status without modifying anything.

  1. Check which artifacts exist for the feature
  2. Determine current pipeline stage
  3. Display a status table showing each phase and its completion state
  4. Highlight the next action needed

validate

Verify artifact integrity and completeness.

  1. Read all artifacts for the feature
  2. Check required sections are present in each artifact
  3. Verify cross-references between artifacts are valid
  4. Report any gaps or inconsistencies

Storage Rules

Artifact Type Storage Path Phase
Problem Spec docs/artifacts/<feature>/problem_spec.md requirements
Design Spec docs/artifacts/<feature>/design_spec.md architecture
Design Review docs/artifacts/<feature>/design_review.md design-review
Implementation Plan docs/artifacts/<feature>/implementation_plan.md impl-planning
Code Review docs/artifacts/<feature>/code_review.md code-review
Verification Report docs/artifacts/<feature>/verification_report.md verify
Risk Assessment docs/artifacts/<feature>/risk_assessment.md risk
Checkpoint State .claude/checkpoints/<feature>/checkpoint.json all
Artifact Digest .claude/checkpoints/<feature>/artifact_digest.md all
Execution Log .claude/checkpoints/<feature>/execution_log.md all

Pipeline Stage Detection

Determine the current stage by checking which artifacts exist:

No artifacts           -> stage: not-started
problem_spec.md        -> stage: requirements-complete
design_spec.md         -> stage: architecture-complete
design_review.md       -> stage: design-review-complete
implementation_plan.md -> stage: impl-planning-complete
(source code exists)   -> stage: implementation-complete
code_review.md         -> stage: code-review-complete
verification_report.md -> stage: verification-complete
risk_assessment.md     -> stage: risk-complete
(PR created)           -> stage: pr-created

Required Sections Validation

problem_spec.md

  • Feature name and summary
  • Layers affected
  • Requirements with priorities (P0/P1/P2)
  • Acceptance criteria (Given/When/Then)
  • Non-goals with rationale
  • Assumptions with risk_if_wrong
  • Edge cases
  • Quality gate results

design_spec.md

  • Component catalog (COMP-IDs with paths)
  • Data flows
  • API contracts
  • Data models
  • ADRs (at least 1)
  • Implementation guidelines
  • Quality gate results

design_review.md

  • Review summary with score
  • Findings by dimension (A through F)
  • Score breakdown table
  • Gate decision (APPROVE / APPROVE_WITH_CONCERNS / REJECT)
  • Recommendations

implementation_plan.md

  • Wave plan (Waves 1-7)
  • File list with actions (NEW/MODIFY)
  • Acceptance criteria mapping
  • Risk register
  • Design review concerns addressed

verification_report.md

  • Requirement coverage matrix
  • Test results summary
  • Edge case verification
  • Domain verification results
  • Overall score
  • Quality gate decision

risk_assessment.md

  • Risk catalog
  • Severity scoring
  • Mitigation strategies
  • Ship recommendation

code_review.md

  • Files reviewed
  • Findings by severity
  • Positives noted
  • Verdict

Checkpoint JSON Schema

{
  "feature": "<feature-name>",
  "current_stage": "<stage>",
  "last_updated": "<ISO timestamp>",
  "artifacts": {
    "problem_spec": { "exists": true, "path": "...", "modified": "..." },
    "design_spec": { "exists": true, "path": "...", "modified": "..." },
    "design_review": { "exists": false },
    "implementation_plan": { "exists": false },
    "code_review": { "exists": false },
    "verification_report": { "exists": false },
    "risk_assessment": { "exists": false }
  },
  "pipeline_history": [
    { "stage": "requirements", "completed_at": "...", "result": "approved" },
    { "stage": "architecture", "completed_at": "...", "result": "complete" }
  ]
}

Status Display Format

Feature: <name>
Current Stage: <stage>
Last Updated: <timestamp>

Pipeline Status:
  [x] Requirements    -> problem_spec.md (approved)
  [x] Architecture    -> design_spec.md (complete)
  [ ] Design Review   -> (next step)
  [ ] Impl Planning
  [ ] Implementation
  [ ] Code Review
  [ ] Verification
  [ ] Risk Assessment
  [ ] PR Creation