Skills
skills/ar9av/obsidian-wiki/openclaw-history-ingest/Gen Agent Trust Hub

openclaw-history-ingest

Pass

Audited by Gen Agent Trust Hub on Apr 24, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill accesses sensitive local file paths to identify the user's workspace and vault. It reads from the .env file and the ~/.openclaw directory, which may contain sensitive configuration and session metadata.
  • Evidence: SKILL.md instructs the agent to read OBSIDIAN_VAULT_PATH and OPENCLAW_HISTORY_PATH from the vault's .env file.
  • Evidence: SKILL.md directs the agent to scan and process files within ~/.openclaw/workspace/ and ~/.openclaw/agents/*/sessions/.
  • [PROMPT_INJECTION]: The skill processes untrusted data in the form of past agent transcripts and memory files, which could contain malicious instructions designed to influence the agent's behavior during the distillation process.
  • Ingestion points: ~/.openclaw/workspace/MEMORY.md, ~/.openclaw/workspace/memory/*.md, and ~/.openclaw/agents/*/sessions/*.jsonl (as described in SKILL.md).
  • Boundary markers: SKILL.md includes instructions to "Summarize; do not quote raw transcripts verbatim" and to apply a "Critical privacy filter," though these are soft constraints.
  • Capability inventory: The skill allows the agent to create and update Markdown files within the Obsidian vault (SKILL.md, Step 6).
  • Sanitization: SKILL.md provides explicit instructions to "Remove API keys, tokens, passwords, credentials" and redact private identifiers during processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 24, 2026, 08:11 PM