game-build
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates by reading and processing local documentation files (such as
docs/mvp-first-draft.mdanddocs/world-lore.md) to guide its code generation process. This creates a surface where instructions embedded in those documentation files could theoretically influence the AI's behavior, although this is the intended operational mode of the skill. - Ingestion points:
docs/mvp-first-draft.md,docs/build-registry.md,docs/world-lore.md,docs/quest-registry.md(read in Phase 0 and Phase 4). - Boundary markers: None explicitly specified to delimit external content from internal instructions.
- Capability inventory: The skill performs file system writes to create source code and update the build registry, and it generates shell commands for the user to execute tests.
- Sanitization: No explicit sanitization or validation of the input documentation content is performed beyond checking for specific sections.
Audit Metadata