claude-codex-workflow
CLAUDE.md1 starsChanged 4 months agoMITView on GitHub
Model gate (check first): only models in Anthropic's Mythos-class tier or above (Claude Fable 5 / Claude Mythos 5 and successors) may drive this workflow. If you are Opus, Sonnet, or Haiku (any version), do not invoke it on your own — tell the user about this gate and wait for instructions. Waived only when the user explicitly asks for the workflow in this session. Apply the repository-root WORKFLOW.md to all development in this directory tree. When working from a…
In this repositoryOne other file it gives its agents.
- codex-dispatchSkill
The file
We split this CLAUDE.md into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Very specific: line 8
Write each task spec to the repository-root tasks/TNNN-slug.md.
- Very specific: line 11
If review fails, append ## Revision feedback (Rn) to the task spec and dispatch it again.
- Distinctive: line 12
If review passes, stage only the reviewed files for that task, inspect the staged diff, and create exactly one commit.
- Distinctive: line 13
Keep tasks, reports, implementation, and all other outputs inside the repository.
- Very specific: line 15
When working from a subdirectory, still use the repository-root tasks/, reports/, and dispatch scripts.
≈ “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
Required Workflow
Model gate (check first): only models in Anthropic's Mythos-class tier or above (Claude Fable 5 / Claude Mythos 5 and successors) may drive this workflow. If you are Opus, Sonnet, or Haiku (any version), do not invoke it on your own — tell the user about this gate and wait for instructions. Waived only when the user explicitly asks for the workflow in this session.
Apply the repository-root WORKFLOW.md to all development in this directory tree.
- Claude only plans, splits tasks, dispatches, and reviews; it never writes implementation code.
- Write each task spec to the repository-root
tasks/TNNN-slug.md. - Dispatch through the root scripts:
.\dispatch.ps1 TNNN-slugon Windows or./dispatch.sh TNNN-slugon macOS/Linux. - Review
reports/TNNN-slug.md, then the Git diff, and re-run the task's verification command. - If review fails, append
## Revision feedback (Rn)to the task spec and dispatch it again. - If review passes, stage only the reviewed files for that task, inspect the staged diff, and create exactly one commit; never use
git add -Ain a dirty worktree. - Keep tasks, reports, implementation, and all other outputs inside the repository.
When working from a subdirectory, still use the repository-root tasks/, reports/, and dispatch scripts.
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.