add-task
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local shell script
create-folders.shlocated in the plugin's root directory to initialize the required folder structure. - [INDIRECT_PROMPT_INJECTION]: The skill ingests raw user input and persists it directly into markdown files within the
.specs/tasks/draft/directory. - Ingestion points: User-provided task descriptions/titles passed as arguments to the skill (SKILL.md).
- Boundary markers: User input is placed under a '## Initial User Prompt' header but lacks specific delimiters or 'ignore' instructions to prevent the agent from obeying instructions embedded in the data during future processing.
- Capability inventory: The skill utilizes the
Writetool to create and save markdown files to the local filesystem. - Sanitization: The skill preserves the 'EXACT user input as provided' without any validation, filtering, or escaping of potentially malicious instructions.
Audit Metadata