context-anchor
Recover from context compaction by scanning memory files and surfacing where you left off. Use when waking up fresh, after compaction, or when you feel lost about what you were doing.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/boscoeuk/context-anchorWhat This Skill Does
The Context Anchor skill is a specialized recovery utility designed for OpenClaw agents to re-establish their mental model after system-level context compaction or session resets. When an agent experiences memory loss due to context window constraints, the Context Anchor scans internal memory files to reconstruct the narrative of the current project. It extracts active task statuses, identifies critical decisions made in recent daily logs, highlights open loops, and lists currently relevant files. By synthesizing information from memory/current-task.md, recent date-stamped logs, and active context directories, it acts as a 'state restore' mechanism that brings the agent up to speed without human intervention.
Installation
Installation is handled via the ClawHub CLI. Ensure your environment meets the dependency requirements (a Unix-like shell with find, grep, and date).
Command:
clawhub install openclaw/skills/skills/boscoeuk/context-anchor
Once installed, you can add it to your AGENTS.md file under an "Every Session" block to automate its execution at the start of every interaction.
Use Cases
- Post-Compaction Recovery: Immediately restore the agent's knowledge after the context window has been cleared or truncated.
- Session Resumption: Reorient the agent when switching between different projects or after long breaks in development.
- Status Reporting: Quickly identify blockers and pending items (open loops) at the start of a meeting or coding session.
- Decision Audit: Verify past architectural decisions to ensure consistency in ongoing work.
Example Prompts
- "I'm back, run the context anchor so we can pick up where we left off yesterday."
- "I feel lost, what were the main blockers and decisions from the last three days?"
- "Show me the current state of my project by anchoring to the latest files and open loops."
Tips & Limitations
- Tip: Use the
--daysflag if your project has a slow burn; increasing the scan to 5 or 7 days can help identify long-standing blockers that might otherwise be missed. - Tip: Ensure your daily logs consistently use prefixes like "Decision:" or "TODO:" to help the script regex parser capture data accurately.
- Limitation: The skill is strictly read-only for local files; it cannot recover information that was not previously saved to the
memory/orcontext/directories. It is only as effective as the logging discipline you maintain in your daily logs.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-boscoeuk-context-anchor": {
"enabled": true,
"auto_update": true
}
}
}Tags(AI)
Flags: file-read