Install from the repository root after purchase. CodexFlow adds AGENTS.md, 2,000+ specialist skills, 11 agents, persistent project memory, 12 safe commands, 9 automated checks, guardrails, and handoffs as plain-text workflow files you can review in Git.
CodexFlow is independent and is not affiliated with, endorsed by, or sponsored by OpenAI.
Before you start
You need:
- A paid Personal or Commercial Solo order
- The signed activation file from your secure access page
- The signed V2 payload manifest and verified archive from your secure access page
- A local code repository
- Git installed
- Node.js and npm available in your terminal
- Terminal access from the repository root
Recommended before running CodexFlow:
git statusCommit or stash important work before installing new workflow files. CodexFlow is designed to create human-readable files, but you should still start from a clean or intentionally prepared working tree.
1. Prepare a clean branch
From the root of the repository you want to structure, install the CLI package used by CodexFlow hooks and local commands:
npm install -D create-codexflowThe exact package source and fulfillment files are provided through your secure order access page.
2. Activate your license
Place the activation file somewhere private on your machine, outside public repositories, then run:
codexflow license activate --activation-file ./codexflow.activation.jsonCheck the license state:
codexflow license statusExpected behavior:
- Personal shows non-commercial rights.
- Commercial Solo shows commercial solo rights.
- Raw license secrets are not displayed in generated premium file watermarks.
3. Install the CodexFlow V2 Individual payload
Install the workflow payload from the files provided through your secure access flow:
codexflow payload install \
--manifest ./codexflow-v2-individual-core-2.0.0.manifest.json \
--archive ./codexflow-v2-individual-core-2.0.0.tar.gzThe CLI checks activation signature, payload manifest signature, tier, entitlement, payload id, archive checksum, and archive size before writing files.
4. Emit workflow files
Generate the Codex operating stack:
codexflow emitOutput is emitted with customer-specific license metadata and watermarks that record tier and commercial-use status without exposing raw license keys.
Expected installed locations include:
AGENTS.md.agents/skills.codexflow/agents.codexflow/memory.codexflow/commands.codexflow/checks
5. Review and verify
Review the generated files before committing them:
git status
git diff
codexflow license status
codexflow payload list
codexflow doctorA healthy setup should show:
- A trusted activation
- The correct tier
- The expected commercial-use flag
- Installed CodexFlow V2 Individual payload status
- 2,000+ specialist skills, 11 agents, 12 safe commands, 9 checks, and memory files available
- Reviewable Git changes
For Personal, commercial use should be false. For Commercial Solo, commercial use should be true.
Commit the baseline once you are satisfied:
git add .
git commit -m "Install CodexFlow V2 operating stack"6. Work with CodexFlow in a real session
A practical CodexFlow session looks like this:
- Start from a clean branch.
- Ask Codex to read
AGENTS.mdand relevant project memory. - Use the appropriate CodexFlow skill, agent role, safe command, or automated check.
- Let Codex propose a plan before editing when risk is high.
- Review risky files before changes are applied.
- Run tests or checks.
- Update memory and handoff notes.
- Commit reviewable work.
CodexFlow does not replace engineering judgment. It gives Codex and humans a shared repo workflow.
7. Removing CodexFlow
To remove CodexFlow-generated files, use the uninstall flow:
codexflow uninstall --yesTo remove an installed payload:
codexflow payload remove codexflow-v2-individual-coreReview the Git diff after uninstalling.