research
Pass
Audited by Gen Agent Trust Hub on Jul 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute the local CLI command
doctrine slice research <id>to initialize research folders and manage state baselines. - [COMMAND_EXECUTION]: The agent is directed to spawn 'research agents' defined by the project's governance documentation. This involves executing arbitrary project-specific commands or models and capturing their stdout to local files.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from external sources into the agent's context.
- Ingestion points: Reads output from external research threads (
research/raw/<thread>.md) and the project's governance document. - Boundary markers: Absent. The skill lacks explicit delimiters or instructions to ignore embedded commands within the ingested research data, although it mentions stripping preambles.
- Capability inventory: The skill facilitates file creation/writing within the
.doctrine/directory and the execution of shell commands via thedoctrineCLI and research agents. - Sanitization: No evidence of sanitization, escaping, or validation of the data received from research agents before it is distilled into the final research artifact.
Audit Metadata