dispatch-authoring
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [NO_CODE]: The skill consists entirely of Markdown documentation and templates. It does not include executable scripts, shell commands, or network operations.\n- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for processing user-provided specifications into agent instructions, which creates an indirect prompt injection surface.\n
- Ingestion points: Untrusted data enters the agent context via the user-provided 'plan or spec' referenced in the
SKILL.mdinstructions.\n - Boundary markers: The skill relies on structural headers rather than explicit delimiters (e.g., XML tags) to isolate the untrusted specification from the dispatch framework.\n
- Capability inventory: The skill generates instructions for coding agents that typically possess file modification, command execution, and network (PR) capabilities.\n
- Sanitization: The workflow does not include specific sanitization or filtering of the source specification, although it encourages strict verification commands to confirm final output.
Audit Metadata