piv-implement
Warn
Audited by Gen Agent Trust Hub on Aug 11, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute "ALL validation commands from the plan in order." Because these commands are defined in an external file passed via the
$ARGUMENTSparameter, the skill allows for the execution of arbitrary shell commands on the user's system.- [REMOTE_CODE_EXECUTION]: The skill serves as a runner for instructions and scripts located in an external plan file. If a user points the skill to a plan file originating from an untrusted remote repository, it facilitates remote code execution.- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests instructions from an external source (the plan file) and explicitly directs the agent to follow them "exactly" and "carefully" without implementing boundary markers or sanitization.\n - Ingestion points: Implementation plan file path provided via
$ARGUMENTS.\n - Boundary markers: None present; the instructions command the agent to follow the tasks in the external plan exactly.\n
- Capability inventory: Full access to shell for validation commands and file system access for implementation.\n
- Sanitization: None; the skill does not validate or filter the commands extracted from the plan file.
Audit Metadata