Openclaw Skill Intelligence Ingestion
ReviewAudited by ClawScan on May 10, 2026.
Overview
The skill is coherent for link analysis, but it automatically reads shared content and writes persistent Obsidian/OpenClaw memory entries without asking first.
Install this only if you want shared URLs and external text to be automatically analyzed and stored in your Obsidian vault and OpenClaw memory. Consider changing the workflow to ask before writing notes or updating MEMORY.md, and periodically review or clean the generated memory entries.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
A casual pasted link or article could trigger browser/web access and create or modify local notes and memory before the user reviews what will be stored.
This combines broad automatic triggers with web reading/searching and local file writes, so a normal shared URL can cause tool use and persistent local changes without confirmation.
execute this pipeline automatically. Do NOT ask for permission — just process it.
Add an approval step before writing files or updating memory, or limit automatic mode to a clearly configured vault and provide an easy opt-out/dry-run mode.
Misleading or prompt-injected external content could be preserved as trusted OpenClaw context and affect future decisions, priorities, or architecture notes.
The skill stores summaries and assessments from external, potentially untrusted content into persistent memory that may influence later agent behavior.
Always append to today's daily log: `~/.openclaw/workspace/memory/YYYY-MM-DD.md`; If strategic value is 🔴 Critical: Also update `MEMORY.md`
Treat ingested external content as untrusted, preserve provenance, require user approval for MEMORY.md changes, and provide review/rollback guidance for memory entries.