project2agentworkos

name	project2agentworkos
description	Use when converting projects, Codex threads, failure reviews, half-finished ideas, README/code/output assets, or repeated workflow problems into AgentWorkOS assets: Agent roles, Memory rules, Skills, MCP/tool needs, Workflows, hard Rules, and Hooks. Also use when selecting an Agent role before a task or integrating distilled content into local Codex.

Project2AgentWorkOS

Project2AgentWorkOS turns real work traces into a reusable AgentWorkOS.

Core Rule

Do not only summarize. Extract future operating capability.

projects + threads + failures + half-finished work
-> Project2AgentWorkOS
-> AgentWorkOS
-> Agent + Memory + Skills + MCP + Workflow + Rules + Hooks

Workflow

1. Select a primary role from the Agent role library before executing.
2. If risk is high, select a verifier role.
3. Inspect the project/thread/file evidence.
4. Classify the work: Active / Watch / Archive / Reference.
5. Extract into the seven AgentWorkOS layers.
6. Create or update a durable artifact: role card, memory, skill, checklist, template, release proof, rule, or hook-ready discipline.
7. Avoid raw private exports. Publish distilled content only.

Role Selection

Use this default mapping:

Task	Primary role	Verifier
Workspace scan	Project Inventory Manager	Quality Reviewer
Thread to memory	Memory Rule Manager	Project Alchemist
Codex integration	Codex Setup Manager	Quality Reviewer
GitHub release	Release Manager	Quality Reviewer
Agent role design	Role Planner	Project Alchemist

Completion Gate

A Project2AgentWorkOS task is incomplete until at least one of these exists:

• a role card
• a memory rule
• a reusable skill
• an MCP/tool requirement
• a workflow/checklist/template
• a hard rule
• a hook candidate or lifecycle check
• public release evidence
• an explicit archive decision

References

• Read references/agent-selection.md when role choice is non-obvious.
• Read references/codex-substrate.md when the task involves local .codex usage or privacy boundaries.