workflow-prd-to-plan
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests potentially untrusted data from local specification files and interpolates them directly into the system prompts of specialized subagents.
- Ingestion points: The skill reads files from
.specs/features/*/prd.mdand.specs/features/*/01-requirements.mdto guide the analysis. - Boundary markers: Untrusted content from external files is inserted into subagent prompts without explicit delimiters or instructions to ignore embedded directives, which could allow a malicious PRD to influence the agent's behavior.
- Capability inventory: The skill possesses file system access (
Glob,Read,Write) and the capability to spawn multiple specialized subagents (code-explorer,backend-architect,principal-software-engineer) to interact with the codebase. - Sanitization: No sanitization, validation, or escaping is performed on the ingested document content before it is interpolated into the executable subagent prompts.
Audit Metadata