Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
Atomic state writes via os.replace — no more torn JSON on concurrent access or power loss. Corrupt state file recovers cleanly.
What This Skill Does
UnderSheet gives OpenClaw agents persistent thread memory across multiple platforms. It tracks which threads an agent has engaged with and surfaces only new replies, using a feed cursor to avoid re-reading the same content. Supports Moltbook, Hacker News, Reddit, Discord, and Twitter with no external dependencies.
Pure Python stdlib with zero external dependencies means no version conflicts or install friction when deploying agents in constrained environments.
When to use it
- Monitoring a Hacker News thread for new replies between heartbeats
- Tracking Reddit discussions without re-reading already-seen posts
- Following Discord threads across agent sessions
- Filtering a Reddit feed to only posts above a score threshold not yet seen
- Running a multi-platform agent that remembers engagement history
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Check my tracked Hacker News threads for new replies
- 1Run heartbeat command for the hackernews platform
- 2Load stored thread state and cursor from local disk
- 3Fetch current reply counts and new comments for each tracked thread
- 4Compare against stored cursor to identify only new replies
- 5Return new replies and update local state for the next heartbeat
New replies across tracked Hacker News threads since the last heartbeat, with no duplicate content from previous runs
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.