@dotdotgod/codex

Change a file, know what else must be checked.

Codex adapter for dotdotgod's context curation workflow. It packages reusable skills that help Codex load curated project memory, plan from docs before implementation, and initialize the shared agent documentation scaffold so changes start with the right specs, tests, plans, and archive map.

What Gets Better?

• Codex can start from AGENTS.md and the dotdotgod docs map.
• Load guidance prefers dotdotgod load-snapshot <root> --json when the CLI is available, then falls back to README-index reads.
• Codex can use docs structure as retrieval intent: specs for behavior, architecture for rationale, tests for verification, plans for current work, and archive indexes for past decisions.
• Product intent, design rationale, and verification standards stay in durable docs.
• Planning work captures current intent in docs/plan/<task-slug>/README.md before implementation.
• Completed plans and temporary reports use the same archive structure as Pi and Claude Code, turning outcomes into future context.
• dd:load, dd:plan, and dd:init can be used as command-like trigger phrases where direct slash commands are unavailable.

Included

• Codex plugin manifest: .codex-plugin/plugin.json
• Skills:
  • project-load: load project memory read-only.
  • doc-first-planning: plan from docs before implementation.
  • project-initializer: initialize shared agent docs and docs folders, using dotdotgod init when available and the bundled fallback when not.

Codex may not expose the same slash-command model as Pi or Claude Code. Treat dd:load, dd:plan, and dd:init as command-like trigger phrases for these skills unless the active Codex plugin runtime provides direct command registration.

Optional Hooks

Codex can run lifecycle hooks from trusted Codex configuration layers. dotdotgod does not require hooks: the bundled skills and dd:load, dd:plan, and dd:init trigger phrases work without them.

Use hooks only when you want opt-in reminders or validation around the same workflow. See hooks/README.md for advisory examples and stricter plan-safety patterns.

Local Development

Run package checks:

pnpm --filter @dotdotgod/codex run verify
pnpm --filter @dotdotgod/codex run pack:dry-run

Shared Contract

• AGENTS.md remains canonical.
• CODEX.md stays thin and points to AGENTS.md.
• Active plans use docs/plan/<task-slug>/README.md.
• Completed plans move to docs/archive/plan/<task-slug>/.
• Temporary reports move to docs/archive/report/<report-slug>/.
• docs/archive/README.md is the archive map; archive bodies should be read only when targeted.

Compared with Graphify-Style Memory

This adapter packages reusable workflow skills. It guides Codex to prefer a bounded dotdotgod load snapshot when available, avoid broad archive scans, and follow README indexes before reading raw files.

The strength is structured retrieval: project docs declare which files are rules, specs, architecture, verification, active intent, or historical memory. That keeps the memory layer portable across Codex runtimes and useful on small tasks where a large graph report would be overhead.