past-conversations

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill introduces a surface for indirect prompt injection by instructing the agent to query and process historical message content.
  • Ingestion points: Past conversation data is ingested from the 'part' table within the local SQLite database at '~/.local/share/opencode/opencode.db'.
  • Boundary markers: The instructions do not define clear delimiters or provide the agent with guidance to ignore any instructions embedded within the retrieved conversation data.
  • Capability inventory: The skill uses the 'sqlite3' command-line utility to perform data retrieval.
  • Sanitization: No sanitization, filtering, or validation logic is present to ensure that the retrieved text from the 'data' field does not contain malicious prompts or control sequences.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 01:47 PM