Memory Ingest
PassAudited by VirusTotal on May 11, 2026.
Overview
Type: OpenClaw Skill Name: memory-ingest Version: 0.1.0 The memory-ingest skill (SKILL.md) is a well-structured tool for converting unstructured text into organized knowledge entities. It follows a safe workflow that includes searching for existing data, performing optional web research, and explicitly requiring user approval before creating new notes via the write_note function. There are no signs of malicious intent, data exfiltration, or unauthorized execution; the instructions focus entirely on the stated purpose of knowledge management.
Findings (0)
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.
The skill can change the user's Basic Memory by adding notes and entities, so incorrect extraction or approval could create inaccurate records.
The skill instructs the agent to create Basic Memory content, but it also includes an approval step before creation, making the write behavior disclosed and user-controlled.
Before creating anything, present what you found and what you'd like to create... Approve all / select individually / skip entity creation?
Review the proposed entities and source-note details before approval, especially for names, relationships, commitments, and dates.
Private meeting details, email contents, or misleading instructions embedded in source text could become long-lived memory entries.
The skill intentionally persists unstructured external content into Basic Memory, which may include sensitive data or untrusted text that future memory retrieval could reuse.
Meeting transcripts, conversation logs, pasted documents, email threads — anything with information worth preserving gets parsed, cross-referenced against existing knowledge, and written as proper notes.
Only ingest material you are comfortable storing in Basic Memory; redact secrets or confidential details and treat pasted source text as data, not instructions.
If the source material is confidential, even searching for extracted entity names or project topics could reveal sensitive associations externally.
The optional research step may send extracted names, organizations, or topics from the pasted material to an external web search provider.
For proposed entities where more context would be valuable, do a brief web search (2-3 queries max per entity).
Skip optional web research for confidential inputs unless the user explicitly approves sharing those search terms externally.