reflect
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes historical message data from previous OpenCode sessions, creating an attack surface for indirect prompt injection if those sessions contained malicious instructions.\n
- Ingestion points: The skill queries the
messagetable from a local SQLite database at/home/mhenke/.local/share/opencode/opencode.db.\n - Boundary markers: Absent. The instructions do not define delimiters or provide guidance to the agent to treat the ingested session data as untrusted.\n
- Capability inventory: The skill can execute shell commands via
bun -eand write summary JSON files to the local filesystem at~/.config/opencode/oh-my-opencode-slim/reflections/sessions/.\n - Sanitization: Absent. No filtering or validation of historical message content is performed before the agent analyzes it for patterns.\n- [COMMAND_EXECUTION]: The skill utilizes
bun -eto execute JavaScript logic for querying the SQLite database and managing its reflection cache.
Audit Metadata