← ClaudeAtlas

agile-vlisted

Use when writing traceable software requirements (REQ IDs, verification criteria), reviewing them for ambiguity, deriving tests from baselined requirements, building a requirements-to-test traceability matrix, or setting up human approval gates.
ymurat/academic-agent-skills · ★ 0 · AI & Automation · score 70
Install: claude install-skill ymurat/academic-agent-skills
# Agile V: requirements and traceability Upstream subset of Agile-V/agile_v_skills (CC-BY-SA-4.0, vetted low risk on 2026-10-31, no scripts): `resources/agile-v/`. 1. Always read `agile-v-core/SKILL.md` first, then the sub-skill that fits. Read `SKILL_ROUTING_GUIDE.md` when unsure. | Sub-skill | Use when | |---|---| | `discovery-analyst` | Turning user research into candidate requirements (OBS/INS lineage IDs) | | `requirement-architect` | Main entry: intent to REQ-XXXX requirements with verification and done criteria, Gate 1, baselining | | `logic-gatekeeper` | Independent review of draft requirements for ambiguity, conflicts, testability | | `test-designer` | Test cases from baselined requirements only, never from code | | `red-team-verifier` | Independent verification against requirements, with approval evidence | | `validation-agent` | Intended-use validation with representative users | | `agile-v-human-oversight` | Blind human precommit and approval-gate evidence (draft status upstream) | | `skills/graph-traceability-agent` | REQ to component to test matrices, orphan detection | 2. Runtime contracts cited by the sub-skills are in `docs/agile-v-runtime/`. 3. References to skills not kept (pipeline, lifecycle, compliance, aibom, gxp): ignore them, or map compliance questions to eu-ai-act, iso42001, nist-ai-rmf, spi-assessment; test design questions stay with test-designer. 4. Human approval gates are real decisions: present them to the user, never approve on their beha