PraxisKit
CLAUDE.md4 starsChanged 7 months agoMITView on GitHub
This repo packages two Claude Code / Codex plugins for the marketplace: Install one or the other, not both (PraxisKit bundles prd-to-kanban).
In this repository2 other files it gives its agents.
- AGENTS.mdAGENTS.md
- prd-to-kanbanSkill
The file
We split this CLAUDE.md into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 47
Test by running the skill against examples/sample-prd.md and inspecting generated work/kanban.md and work/SUBAGENT.md.
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Overview
This repo packages two Claude Code / Codex plugins for the marketplace:
- prd-to-kanban (standalone) — converts PRDs/requirements into
work/kanban.mdandwork/SUBAGENT.md. Planning-only; does not execute tasks. - praxiskit (full pipeline) — typed transform chain:
seed-to-idea -> idea-to-prd -> prd-to-task-graph -> task-graph-to-batch -> batch-to-build -> review-and-accept, withnext-iterationas the driver skill (default mode: one step; budget mode: bounded loop).
Install one or the other, not both (PraxisKit bundles prd-to-kanban).
File Structure
SKILL.md # Standalone prd-to-kanban skill entry (thin wrapper)
plugins/prd-to-kanban/SKILL.md # prd-to-kanban source of truth
plugins/prd-to-kanban/skills/prd-to-kanban/ # Plugin-packaged entry (thin wrapper)
plugins/prd-to-kanban/.claude-plugin/ # Claude Code plugin manifest
plugins/prd-to-kanban/.codex-plugin/ # Codex plugin manifest
plugins/praxiskit/skills/seed-to-idea/ # PraxisKit planning transform
plugins/praxiskit/skills/idea-to-prd/ # PraxisKit planning transform
plugins/praxiskit/skills/seed-to-task-graph/ # Light recipe graph transform
plugins/praxiskit/skills/prd-to-task-graph/ # PRD to task graph transform
plugins/praxiskit/skills/task-graph-to-batch/ # Batch scheduling transform
plugins/praxiskit/skills/batch-to-build/ # Authorized execution transform
plugins/praxiskit/skills/review-and-accept/ # Review packet + acceptance + archive (merged)
plugins/praxiskit/skills/next-iteration/ # Recursive driver (single-step + budget mode)
plugins/praxiskit/RECIPES.md # Consolidated recipe reference
plugins/praxiskit/.claude-plugin/ # PraxisKit Claude Code plugin manifest
plugins/praxiskit/.codex-plugin/ # PraxisKit Codex plugin manifest
.claude-plugin/marketplace.json # Claude Code marketplace catalog
.agents/plugins/marketplace.json # Codex marketplace catalog
examples/ # Sample PRD input and kanban output
CLAUDE.md # This file
AGENTS.md # Repo context for Codex
README.md # User-facing install/usage docs
CHANGELOG.md # Version historyModifying Skills
- prd-to-kanban behavior: edit
plugins/prd-to-kanban/SKILL.md— the single source of truth. Keep rootSKILL.mdandplugins/prd-to-kanban/skills/prd-to-kanban/SKILL.mdas thin wrappers. - PraxisKit steps: edit the respective
plugins/praxiskit/skills/<step>/SKILL.mddirectly. Each step is self-contained. - Plugin manifests:
interfaceblock in.claude-plugin/plugin.jsoncontrols marketplace presentation. - Test by running the skill against
examples/sample-prd.mdand inspecting generatedwork/kanban.mdandwork/SUBAGENT.md.
Key Design Constraints
- Two output files:
kanban.md(dynamic status tracker) +SUBAGENT.md(static shared context, under 80 lines). Orchestrator passesSUBAGENT.mdto every subagent — subagents read it themselves. - Agents edit, humans read: The board is a living document updated by orchestrating agents during execution
- Topological layering drives parallelism: Same layer + disjoint write scopes = safe to parallelize
- Frozen interfaces first: Schema/contract tasks are always Wave 0
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

