oma-recap
Pass
Audited by Gen Agent Trust Hub on May 3, 2026
Risk Level: SAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: Accesses the local file
~/.claude/history.jsonl, which contains sensitive private conversation logs. This access is essential for the skill's primary function of generating work summaries. All data processing is performed locally with no external network transmission identified. - [COMMAND_EXECUTION]: Executes the
omaCLI (vendor tool) and thejqutility to fetch and normalize conversation data. - [PROMPT_INJECTION]: Features an indirect prompt injection surface due to the analysis of untrusted data from AI conversation histories. 1. Ingestion points:
~/.claude/history.jsonl. 2. Boundary markers: Absent. 3. Capability inventory: Local file writing in the.agents/directory. 4. Sanitization: No sanitization or filtering is applied to the ingested history content.
Audit Metadata