takt-task-builder
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a bundled shell script 'scripts/validate-order-md.sh' to verify the structure of generated task specifications. It also instructs the agent to perform validation using a script from a related skill ('takt-piece-builder') when editing piece configurations.
- [PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection. It collects task requirements from users or GitHub Issues and places them directly into 'order.md' files. This content is eventually processed by the TAKT agent, which possesses capabilities such as shell command execution ('Bash') and file modification ('Edit', 'Write'). 1. Ingestion points: User input for task requirements and content fetched from GitHub Issues. 2. Boundary markers: None present in the 'order.md' templates or the generated files to delimit untrusted content. 3. Capability inventory: The downstream TAKT agent has access to tools including 'Bash', 'Edit', 'Write', 'Read', 'Grep', and 'Glob' as documented in the tool's reference materials. 4. Sanitization: No explicit sanitization of the user-provided or Issue-sourced task descriptions is performed before file generation.
Audit Metadata