fusion-code-conventions
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates as a code-review orchestrator that ingests external file content and possesses write permissions, creating a potential attack surface for indirect prompt injection.
- Ingestion points: The agent reads user-provided files, source code (
.ts,.tsx,.cs,.md), local project configurations (tsconfig.json,biome.json,.editorconfig), and architectural decision records (docs/adr/,CONTRIBUTING.md). - Boundary markers: Absent. There are no specified boundary markers or strict delimiters defined to insulate the source code contents from the agent's prompt context.
- Capability inventory: The skill is authorized to perform project file discovery, execute
git CLIchecks for repository history, and modify files using workspace tools upon user confirmation (SKILL.md). - Sanitization: Absent. No filtering, escaping, or explicit structural validation is performed on the code inputs to strip embedded instructions before parsing.
Audit Metadata