fluff-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill allows the agent to forward user-provided flags directly to a shell script (
larch.sh) using the Bash tool. This design presents a potential argument injection vector if the agent does not strictly validate the flags before execution, potentially allowing a user to specify sensitive output paths or unintended operational parameters.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external run logs and session caches to generate its reports. This creates an attack surface where malicious content within those logs could attempt to influence the agent's behavior when it interprets or relays the report.\n - Ingestion points: The skill reads from the
larch-logs/implement/directory and the~/.cache/larch/sessionsdirectory.\n - Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the tool's output as untrusted or to ignore embedded instructions.\n
- Capability inventory: The skill uses the Bash tool to run the analyzer and the Read tool to access project files.\n
- Sanitization: The skill does not specify any sanitization or filtering of the log content before it is processed by the analyzer or relayed to the user by the agent.
Audit Metadata