implement-with-remote-feedback
Pass
Audited by Gen Agent Trust Hub on Apr 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes various Git commands (
git checkout,git push,git status,git pull) and build system commands (typicallymake) to perform its tasks. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the plan documents it processes.
- Ingestion points: The agent reads external plan files from
docs/plans/plan_*.mdand other referenced materials. - Boundary markers: There are no explicit instructions to ignore or sanitize natural language instructions found within the plan documents; rather, the skill is explicitly told to "FOLLOW THE PLAN" as the source of truth.
- Capability inventory: The skill has the capability to write files, commit/push to remote repositories, and execute shell commands via
maketargets. - Sanitization: There is no evidence of sanitization or validation of the contents of the plan documents before they are used to drive the agent's actions.
Audit Metadata