X-Bookmarks
github.com/sharbelxyz/x-bookmarks- Category
- AI Agents
- Rank
- No. 1109Tools index
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Workflow Automation
- Interfaces
- Agent Skill / Plugin
- Builder
- @sharbelxyz
- GitHub
- 289 stars
- Date
About
An OpenClaw skill that transforms your X/Twitter bookmarks into actionable AI agent tasks. Instead of letting bookmarks pile up unused, it categorizes them and proposes specific actions your AI agent can execute.
What it does
X-Bookmarks gives an AI agent two ways to retrieve saved posts: browser-cookie access through bird, or OAuth access through X API v2. Both routes produce a common JSON shape. Agent instructions then define digests, topic grouping, stale-item cleanup, history analysis, and suggested follow-up work.
Why it's ranked here
The dual retrieval design is practical, and normalizing both sources keeps later workflows independent of authentication choice. The important caveat is architectural: the repository implements fetching and authentication, while categorization, pattern finding, and action proposals remain agent instructions rather than deterministic application code.
What's good
OAuth uses PKCE, validates callback state, stores tokens with restricted file permissions, and attempts refresh automatically. The API fetcher handles pagination, media, authors, quoted posts, metrics, bearer-token overrides, and rate-limit messaging. The bird route adds browser profiles, thread context, folders, and unbookmarking.
Tradeoffs
Neither access route is frictionless. Bird requires Node.js, a separate global package, and usable browser cookies. The API route requires Python 3.10+, a developer account, app configuration, and OAuth authorization. Scheduled digests require an external cron setup. No supplied test suite demonstrates authentication, normalization, or pagination behavior.
How to use it well
Use it when an OpenClaw agent already handles recurring personal research and can turn saved posts into concrete follow-up work. Start with a small digest, then use full history for pattern analysis. It does not replace a scheduler, general bookmark manager, or independent analysis application.
Technical notes+
SKILL.md contains the OpenClaw behavior, backend selection rules, digest formats, cleanup logic, and cron guidance. scripts/fetch_bookmarks.sh is a thin Bash wrapper around bird CLI. scripts/x_api_auth.py implements OAuth 2.0 PKCE with a localhost callback, local configuration, token persistence, and refresh support. scripts/fetch_bookmarks_api.py calls X API v2, paginates results, and normalizes tweets into the bird-compatible JSON structure. references/auth-setup.md documents Chrome, Firefox, Brave, manual-cookie, and developer-app setup. The supplied repository tree shows no automated tests.
Observed
- License
- MIT
- Implementation languages
- Python and Bash
- Install surface
- OpenClaw skill folder, installed by copying it into the skills directory
- Interfaces
- OpenClaw natural-language skill plus command-line fetch and authentication scripts
- Authentication backends
- bird CLI browser cookies or X API v2 OAuth 2.0 PKCE
- Runtime requirements
- Node.js and npm for bird, or Python 3.10+ and an X Developer account for API access
- Testing structure
- No test directory or automated test files appear in the supplied repository tree
Read from README.md, scripts/x_api_auth.py, scripts/fetch_bookmarks_api.py, SKILL.md, references/auth-setup.md, scripts/fetch_bookmarks.sh.
What it can do
Fetch X/Twitter bookmarks
X/Twitter account credentials → Collection of bookmarked tweets and content
Categorize bookmarks automatically
Raw bookmark data → Organized bookmark categories
Generate actionable tasks from bookmarks
Categorized bookmarks → Specific action proposals for AI agents
Transform passive bookmarks into executable tasks
Stored Twitter/X bookmarks → AI agent-ready task instructions
Propose specific actions for bookmarked content
Individual bookmarked tweets or threads → Concrete action recommendations
Tags
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.