brownfield-drift
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external project files (
PLAN.md) to define architectural constraints, creating a surface for indirect instructions. - Ingestion points: The
wednesday-skills driftcommand (referenced in SKILL.md) reads user-defined boundaries and rules fromPLAN.md. - Boundary markers: There are no explicit instructions or delimiters provided to the agent to distinguish between data and potentially malicious instructions embedded in the processed configuration files.
- Capability inventory: The skill is authorized to execute
wednesday-skills driftandgit logcommands using the Bash tool. - Sanitization: The instructions do not define sanitization or validation steps for the content parsed from
PLAN.mdbefore it is interpreted by the agent. - [COMMAND_EXECUTION]: The skill executes specific shell commands through the Bash tool.
- Evidence: The skill is granted permissions for
Bash(wednesday-skills drift *)andBash(git log *). These are scoped to the vendor's own utility and standard git operations.
Audit Metadata