stelow — your agentic product team. AI agents less like coding assistants, more like cross-functional product team.
This repository also configures its own agents. See what stelow tells them →
Skill Claude CodeCodex
Run post-implementation testing protocol. Triggers when: user says 'test this', 'run tests', 'QA', 'dogfood', 'check quality', user finishes implementing a feature, or when a PR is ready for review. Also triggers on mentions of: test coverage, accessibility audit, WCAG, design review, code review, subagent review, UX…
Skill Claude CodeCodex needs its repo
Standalone awareness: inside stelow, quality gates critique depth (experimental → static a11y baseline, production → live site). Standalone defaults to production quality (live-site mode when a URL exists, else codebase/browserless mode, 80% coverage). Works with URL, source directory, or screenshot — no stelow…
Skill Claude CodeCodex
This skill is the first thing a workflow-mode stelow session loads (via the marker STELOWWORKFLOW=1). It does not run stage logic. Its only job is to bootstrap the workflow: classify intent, write state.md, and pick the first stage.
Skill Claude CodeCodex
The router is the workflow control plane. It runs once per stage transition. It does not implement any stage logic itself; it only moves state forward and loads the next stage skill.
Skill Claude Code
Use this skill to turn an approved product proposal, technical plan, or codebase context into a small set of coherent delivery scopes.
Skill Claude CodeCodex
Use this skill during the existing Interface stage. It does not create a new root stage and it does not own card state or execution lifecycle.
Skill Claude CodeCodex needs its repo
This skill executes the Architecture Alternatives phase: diverging construction directions before any commitment. It is the structural sibling of interaction alternatives. Interaction asks how it behaves. Architecture asks how it is built.
Skill Claude CodeCodex needs its repo
Use this skill during the existing Architecture stage. It does not create a new root stage and it does not own card state or execution lifecycle.
Skill Claude CodeCodex
Two stages of the short-cycle method. Stage 1 explores the problem space; stage 2 turns what you learned into a market you can act on. Both come before any solution thinking — if you skip them you will build for an audience you invented.
Skill Claude CodeCodex
Objective: go beyond curiosity and test real commitment. Ask people to invest something valuable: their time or their money. This is what separates curious people from true early adopters.
Skill Claude CodeCodex
Objective: test the execution and usability risk — can you actually deliver the solution? Instead of jumping to full development, experiment with a "valuable manual process".
Skill Claude CodeCodex
Objective: leave the field of ideas and create a tangible offer. Test the value risk: do people show genuine interest when presented with the proposed solution?
Skill Claude CodeCodex
Objective: explore and prioritize possible paths, balancing what is desirable for the audience with what is viable and strategic for the business.
Skill Claude CodeCodex
Objective: turn accumulated learning into a sustainable business. Once the problem, the solution and the delivery capability are confirmed, the focus becomes business risk.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: