roadmap-recommend
Pass
Audited by Gen Agent Trust Hub on May 29, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by reading content from the project's filesystem (specifically files in the
.spec-driven/directory) and incorporating it into the agent's reasoning. - Ingestion points: The skill reads
.spec-driven/config.yaml, roadmap INDEX.md, milestone files, and specification files (SKILL.md, Step 1). - Boundary markers: There are no explicit instructions to the agent to isolate or treat ingested file content as untrusted or to ignore any embedded instructions.
- Capability inventory: The skill executes CLI commands via
node {{SKILL_DIR}}/scripts/spec-driven.jsand performs multiple file writes to createproposal.md,design.md,tasks.md,questions.md, and delta spec files (SKILL.md, Step 6 and 7). - Sanitization: No evidence of sanitization or validation of the input data before it is used for file generation or reasoning.
- [COMMAND_EXECUTION]: The skill relies on executing local JavaScript files via Node.js (
spec-driven.js) to manage project state, scaffold changes, and verify artifacts. While these are local project scripts, they constitute the skill's primary mechanism for state modification and system interaction.
Audit Metadata