{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreichpz74skceaoowb2ktbakx5vzzjzkp6sgmx7xaw53eixywwbszli",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mpifnjcayi72"
  },
  "path": "/t/does-chatgpt-apps-support-reading-standard-mcp-resources/1384027#post_3",
  "publishedAt": "2026-06-30T06:23:22.000Z",
  "site": "https://community.openai.com",
  "textContent": "I would not assume general MCP resources are model-readable in Apps until the platform exposes a clear resources/list + resources/read path for that surface. For now I’d keep the tool descriptions tiny but add an explicit tool like get_workflow_guidance(scope) that returns the shared rules by key. That keeps the schema small, makes the read auditable, and avoids duplicating the same guidance into every tool. Resource links can still be useful for UI or human inspection, but I would not make core agent behavior depend on them being discovered implicitly.",
  "title": "Does ChatGPT Apps support reading standard MCP resources?"
}