daemon-logs
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads and displays content from log files (
~/.asciinema/logs/chunker.log) that may incorporate data from external or untrusted sources. - Ingestion points: Log data is read via
tailandgrepcommands inSKILL.mdand presented to the agent. - Boundary markers: The output uses basic text headers like
=== Daemon Log ===but does not include explicit warnings to the agent to ignore potential instructions embedded within the logs. - Capability inventory: The skill utilizes the
Bashtool to perform file operations and text processing. - Sanitization: There is no evidence of sanitization or escaping of the log content before it is displayed to the agent.
Audit Metadata