{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreifukqnzs2zf5grpmh2eyyvf2zlluqcv4vyqvtpvzvvi43hq45wb7q",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mopekbrak2x2"
  },
  "path": "/t/add-persistent-user-preference-recall-across-codex-cli-conversations/1378787#post_13",
  "publishedAt": "2026-06-20T06:30:55.000Z",
  "site": "https://community.openai.com",
  "textContent": "Agreed that a lot of this is covered once you map the Codex surfaces correctly. The docs make the split clearer than the mental model people often start with:\n\n  * `AGENTS.md` for durable instructions, globally or per repo\n  * skills for reusable workflows that should trigger on a task\n  * MCP for external memory/tools/context\n  * memories for lightweight carried-forward context, where enabled\n  * snippets/custom saved prompts for repeated phrasing or commands\n\n\n\nThe nuance I’d still preserve is that these solve slightly different problems. A skill is great when there’s a recognisable workflow. `AGENTS.md` is better for baseline behaviour: review tone, verbosity, repo conventions, preferred commands, etc. MCP or a “second brain” setup is better when the agent needs to retrieve external context rather than carry it in the prompt.\n\nSo I think the practical answer is not “one memory layer,” but using the right scope: global Codex instructions for how I work, repo `AGENTS.md` for how this codebase works, skills for repeatable workflows, and MCP/memory for recall. This gets pretty close to the precedence model people are asking for, without needing an umbrella project that mixes personal defaults with repo context.",
  "title": "Add persistent user preference recall across Codex CLI conversations"
}