Production-grade skill library + agents + workflows + governance for AI coding agents — full SDLC
Plugin Claude Code
Praxis — production-grade skill library + agents + workflows + governance for AI-augmented software delivery. 91 SKILLs, 17 role agents, 9 workflows, governance gates, project memory taxonomy.
Command Claude Code
Phase B — Architecture + Challenger review. Solution Architect runs; ends at architecturesignoff gate.
Command Claude Code
Brownfield kickoff: comprehension + architecture reconciliation + debt audit + impact analysis.
Command Claude Code
Phase A — Discovery + requirements + NFRs. Product Manager runs; ends at requirementsfreeze gate.
Command Claude Code
Start or resume autonomous drive — iterate the drive protocol (skills/autonomous-drive) against the active task ledger, unattended or in-session, honoring the non-negotiable stops.
Command Claude Code
Record a rich factory-metrics observation for a praxis artifact (skill, agent, workflow, command). Use when you've just finished using an experimental skill and want to capture what worked, what didn't, and what should change before steward review.
Command Claude Code
Single entry point for any new requirement in steady state (story, epic, ticket, change request). Delegates to delivery-lead, which runs requirements-intake to triage, right-size, and route — so you never pick a workflow by hand.
Command Claude Code
Run a configurable creator/reviewer/enhancer loop over an ideation artifact. Harnesses are runtime bindings, so creator/reviewer can be Claude, Codex, Gemini, Cursor, OpenCode, or custom adapters.
Command Claude Code
Production release: assemble evidence, route through productiongolive gate, deploy.
Command Claude Code
On-demand, closed-loop review of an architecture-significant artifact (contracts, roadmap, ADRs, or a path) produced outside a gate. Runs Challenger + Code Reviewer + Security Reviewer, then routes findings back to the owning agent to fix and re-reviews until clear.
Command Claude Code
Run one implementation slice end-to-end via implementation-slice.yaml workflow. ALWAYS delegates to delivery-lead; never spawns specialists directly.
Command Claude Code
Bootstrap a new project: run delivery-planner, set the charter, establish the architecture-doc skeleton.
Command Claude Code
Quarterly library review: factory-evaluation + System Steward + stewardpromotion gate.
Instructions file GitHub Copilot
Instructions for jeet129/praxis, covering github copilot — praxis integration, discovering the right skill, activation flags, where things live and key disciplines (always apply).
Instructions file Gemini CLI
Instructions for jeet129/praxis, covering gemini routing for praxis, where things live, how to discover the right skill, project memory and activation flags.
Agent
The adversarial-review role that fires after the Solution Architect produces a design and before the architecturesignoff gate. Distinct from the SA in objectives, not just context — the Challenger is prompted to PROVE THE DESIGN WRONG. Ships with five selectable sub-personas (scale, security, cost, operations…
Agent
The backend specialist — implements server-side slices against the active backend stack (Java/Spring, Node/TS, or Python). Consumes engineering-standards, the active stack pack (stack-java-spring / stack-node-ts / stack-python), secure-coding, testing-strategy, observability. Produces production-grade code with…
Agent
The pre-merge quality gate. Runs the seven-dimension review against engineering-standards + active stack pack + security + NFR impact + test sufficiency + API/data contracts + documentation. Produces a severity-tagged report (blocker/major/minor/nit) with diff-anchored locations and concrete fix suggestions. Distinct…
Agent
The specialist who owns the data plane. Activated only on engagements with non-trivial data workloads (per delivery-planner's hasdataplane flag). Consumes data-pipeline, data-warehouse-modeling, data-quality, data-governance, plus data-modeling (transactional schemas), secure-coding (data-handling), compliance-privacy…
Agent
The orchestrator persona that runs the Praxis's lifecycle. Loads workflows, evaluates Decision Nodes, routes work to phase-lead and specialist agents, enforces gates per the governance matrix, manages slice transitions, and serializes writes to project memory. Use as the entry-point agent for any project work — every…
Agent
The frontend specialist — implements user-facing slices against the active frontend stack (React + Next.js, Angular, or Vue + Nuxt per frontend-architecture's choice). Consumes engineering-standards, stack-web-frontend + its framework reference, frontend-architecture, design-system, accessibility, secure-coding (FE…
Agent
The implementation phase lead. Owns slice-level planning, task decomposition across FE/BE/Data/ML-AI specialists, dependency management between their work, and slice-level hand-off coordination. NOT responsible for architecture (Solution Architect), product (PM), platform (Platform/SRE), or individual code quality…
Agent
The specialist who owns the model plane AND the agentic AI plane. Activated only on engagements with ML or LLM/agent workloads (per delivery-planner's hasml = true OR hasagenticai = true flags). Consumes the full ML/MLOps family (ml-problem-framing → ml-monitoring-drift + responsible-ai) and the agentic AI family…
Agent
The mobile specialist — implements user-facing slices as a native mobile app against the active mobile stack (stack-flutter). Consumes engineering-standards, stack-flutter, frontend-design (visual craft — the same anti-generic discipline as web), api-design (client-side contract consumption), accessibility…
At most 3 mods per repository are shown here — the rest are on their repository pages: