honey-debt
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands (
grepandgit blame) to retrieve data from the local filesystem. - Evidence: Usage of
grep -rnE '(#|//|--|/*) ?honey:' .andgit blame -L<line>,<line> -- <file>as defined in the Scan and Output sections of SKILL.md. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the codebase (comments starting with
honey:) and incorporates them into its output ledger. Malicious instructions placed in these comments could theoretically attempt to influence the agent's behavior during report generation. - Ingestion points: File contents scanned via
grepinSKILL.md. - Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the comment content as data rather than instructions.
- Capability inventory: The skill is restricted to file reading and reporting; it does not have capabilities for network operations or file writes.
- Sanitization: No sanitization or escaping of the harvested comment strings is specified.
Audit Metadata