Agile v pipeline
🔬 Verifiable AI-Augmented Engineering Framework - Stop AI hallucinations with formal traceability (REQ→ART→TC). Agent Skills for Claude Code, Cursor, VS Code & Copilot. Enterprise-grade: ISO 9001, ISO 27001, GxP-ready. Red Team verification, multi-cycle lifecycle, behavioral anti-patterns.
npx -y skills add Agile-V/agile_v_skills --skill agile-v-pipelineAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
What its author says it does
Copied from the file, not written here
Orchestration pipeline, wave execution, handoff protocols, and checkpoint types for the Agile V 5-stage workflow. Load when orchestrating multi-agent pipelines or managing stage transitions.
The file declares its own license as CC-BY-SA-4.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
2.8 KB, as published. Nobody here has run it
Instructions
Orchestration pipeline for Agile V. Requires agile-v-core loaded first.
Pipeline
Stage 1: draft persisted -> Stage 2: independent findings -> Stage 1: architect revisions -> [Human Gate 1] -> approved/baselined -> Stage 3: Synthesis (Build Agent || Test Designer) -> Stage 4: Verification -> [Human Gate 2] -> Stage 5: Acceptance
Compliance Auditor observes all stages.
Handoffs
- Req Architect persists draft
REQUIREMENTS.md-> Logic Gatekeeper reads without editing. - Gatekeeper records independent findings -> Req Architect revises -> Gate 1 (findings + revision evidence, Human approves).
- Approval creates a baseline; Build Agent || Test Designer use only that baselined revision, with no shared context.
- Build Manifest + Test Cases -> Red Team Verifier.
- Validation Summary -> Gate 2.
Gate 2 prereqs (Phase 1): EVAL_RESULTS.md + VALIDATION_SUMMARY.md EvalGate line. Gate pause (Phase 2): append CHECKPOINTS.md (PENDING + resume_token); resume only with matching APPROVALS.md entry (see agile-v-core).
Stage Failure Handling
If a stage fails (e.g., Logic Gatekeeper rejects requirements, Red Team finds CRITICAL issues), the pipeline stops at that stage. Do not proceed to the next stage.
| Failure | Action |
|---|---|
| Gate 1 rejected | Return to architect revisions; preserve rejected revision and findings |
| Build Agent HALT | Human resolves blocker; Build Agent resumes; do not skip to Stage 4 |
| Red Team FAIL | Return to Stage 3; Builder addresses findings; re-verify affected only |
| Gate 2 blocked | Do not proceed to Stage 5; resolve EVAL gate or get WAIVED approval |
Never advance to Stage 5 with open CRITICAL or MAJOR findings.
Wave Execution
Dependency analysis -> Wave assignment (no-deps = Wave 1) -> Parallel within waves (fresh context each) -> Sequential across waves -> Prefer vertical slices (feature > layer).
Checkpoint Types
| Type | Action |
|---|---|
| Auto | Proceed |
| Human-Verify | Confirm output |
| Human-Decision | Choose alternative |
| Human-Action | Physical/external step |
All except Auto require Human Gate protocol.