Agent Claude Code
Audits coherence and completeness of the GSE-One methodology repository across spec, design, and implementation layers. Activated by /gse-meta-audit. Local to the gensem repo (.claude/agents/), not part of the distributed plugin.
Command Claude Code
Post-audit fix session for the GSE-One methodology repository. Consumes the canonical registry LOCAL/audit/audit.json produced by /gse-meta-audit: verifies findings against LLM fabrication (per-cluster verification pass), persists verdicts in place, applies CONFIRMED fixes cluster by cluster, and closes with the full…
Command Claude Code
Audit the GSE-One methodology repository for coherence (Categories A-D), strategic quality (Category E), and distribution hygiene (Category F). Orchestrates the Python deterministic engine (audit.py, including the 5 Category F distribution jobs) + 23 parallel LLM sub-agents — 28 jobs total, defined in…
Command Claude Code
Run the mandatory GSE-One build pipeline for a commit to main: VERSION bump, CHANGELOG entry, generator --verify, add, commit, push, and the vX.Y.Z tag that triggers the release workflow. Thin executable checklist over CLAUDE.md > Critical rules > Build pipeline (canonical).
Settings file Claude Code
Agent settings declaring 5 allowed tools.
Instructions file
Claude Code instructions for nicolasguelfi/gensem, covering gse-one development guide, project overview, repository structure, agent archetypes and critical rules.
Plugin Claude Code
Structured SDLC management with 22 commands. Adaptive risk analysis, unified backlog, knowledge transfer, worktree isolation, AI integrity safeguards.
Plugin Claude Code
GSE-One — AI engineering companion for structured SDLC management. 24 commands, adaptive risk analysis, unified backlog, knowledge transfer, worktree isolation.
Agent
Evaluates architecture decisions for quality, scalability, and maintainability. Activated during /gse:design and /gse:review.
Agent
Observes the AI+user collaboration across 8 axes (pedagogy + 7 workflow axes) and returns either 5-option P14 preambles (pedagogy) or Inform-tier workflow advice. Invoked by the orchestrator at activity start (pedagogy) and at workflow-monitoring moments (/gse:go, /gse:pause, /gse:compound, sprint promotion, detection…
Agent
Reviews code for quality, security, and maintainability. Activated during /gse:review.
Agent
Manages Hetzner/Coolify deployment lifecycle. Governs safety (credential handling, no-exposure), idempotence (phase tracking), and cost-confirmation gates. Activated during /gse:deploy.
Agent
Challenges the agent's own productions for AI integrity (P16). Activated during /gse:review to hunt hallucinations and challenge assumptions.
Agent
GSE-One main orchestrator agent. Manages the full software development lifecycle with 24 commands under the /gse: prefix. Adapts language, decisions, and autonomy to the user's profile.
Agent
Canonical rule reference for guardrail compliance (Compliance archetype). Defines the Soft/Hard/Emergency tier taxonomy + GUARD-NNN output format. Not spawned at runtime — consulted by activity authors and enforced by system hooks + inline Step 0 preflights.
Agent
Ensures requirements are complete, testable, and traceable. Activated during /gse:reqs and /gse:review.
Agent
Identifies security vulnerabilities and risks. Activated during /gse:review and /gse:design.
Agent
Ensures test coverage, strategy, and evidence quality. Activated during /gse:tests (STRATEGY and TST-SPEC tiers) and /gse:review (IMPL tier). /gse:produce does not invoke this agent — the code-vs-tests relationship is evaluated at REVIEW per spec §6.5.
Agent
Evaluates user experience and accessibility. Activated during /gse:preview and /gse:review.
Hook
Runs before the agent uses a tool for Bash tool calls, running an inline shell check (2 commands). From nicolasguelfi/gensem.
Hook
Runs after a tool call finishes for Bash, Edit and Write tool calls, executing dashboard.py via python3 (3 commands). From nicolasguelfi/gensem.
Skill Claude CodeCodex
Evaluate artefact status against project goals. Identifies covered, partial, and uncovered goals.
Skill Claude CodeCodex
Audit the project for methodology drift. Detects format inconsistencies, missing evidence, skipped steps, and other deviations from the canonical methodology. Triggered manually by /gse:audit or auto-triggered by the orchestrator when drift signals accumulate (Phase 3).
Skill Claude CodeCodex
View and manage project backlog. Triggered when user asks about tasks, work items, what's left to do.