superplan
Pass
Audited by Gen Agent Trust Hub on Mar 31, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests and processes untrusted data from external sources.
- Ingestion points: User-provided feature requirements, stories, and external ticket links (Jira, Linear, GitHub) fetched via MCP as described in Phase 1 (INTAKE) of SKILL.md.
- Boundary markers: While the skill summarizes input, it lacks explicit delimiters or instructions to ignore potential instructions embedded within the ingested requirement text.
- Capability inventory: The skill possesses extensive capabilities, including filesystem read/write access, network searching, sub-agent spawning, and tool execution (e.g., /superbuild, /compact).
- Sanitization: There is no evidence of input validation or sanitization of the requirement data before it is incorporated into the implementation plans.
- [EXTERNAL_DOWNLOADS]: The skill performs targeted web searches for technical research and best practices based on the detected technology stack (Phase 4: RESEARCH).
- [COMMAND_EXECUTION]: The skill uses local commands and platform tools (such as /superbuild and /compact) to manage context, scan the codebase for technology detection (Phase 2: DETECT), and execute the generated implementation phases (Phase 8 and references/SUBAGENT-SYNTAX.md).
Audit Metadata