sdd-tasks

Pass

Audited by Gen Agent Trust Hub on Apr 8, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill performs extensive reading of external project files such as design documents, specifications, and context files (ai-context/, CLAUDE.md, design.md). This creates an indirect prompt injection surface where instructions embedded in these files could influence the agent to generate malicious or incorrect tasks in the final plan.
  • Ingestion points: Multiple files including ai-context/, CLAUDE.md, openspec/specs/, design.md, and proposal.md.
  • Boundary markers: Absent; the skill does not use delimiters or instructions to ignore embedded commands in the source files.
  • Capability inventory: File reading and writing to tasks.md; no direct command execution capabilities are utilized.
  • Sanitization: None; the agent is instructed to treat the loaded content as an authoritative behavioral contract or enrichment.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 8, 2026, 05:45 AM