{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreigkmhzbqolc3lezfduzvw7zwslwkrcvj6sddcqar2iduq25xlx774",
"uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mku2cuxl3l62"
},
"path": "/t/built-a-local-sqlite-task-queue-for-codex-cli-workflows/1380185#post_1",
"publishedAt": "2026-05-02T06:20:13.000Z",
"site": "https://community.openai.com",
"tags": [
"github.com",
"GitHub - daniel-aranda/Agents-todo: A tiny repo-local SQLite task queue for Codex CLI..."
],
"textContent": "I’m experimenting with a local-first task queue for Codex CLI.\n\ngithub.com\n\n### GitHub - daniel-aranda/Agents-todo: A tiny repo-local SQLite task queue for Codex CLI...\n\nA tiny repo-local SQLite task queue for Codex CLI agents.\n\nThe core question: should coding agents use human PM tools like Linear/Jira, or should they have small local queues with claim/lease/review semantics?\n\nI built cxq as one answer: repo-local SQLite, Bash CLI, no server.\n\nCurious if others are solving this with TODO.md, GitHub issues, AGENTS.md conventions, or custom scripts.",
"title": "Built a local SQLite task queue for Codex CLI workflows"
}