ns-harness-bootstrap-brownfield
Pass
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted metadata from external codebases (such as module names, configuration keys, and file structures) to generate reports used by other agents, creating a surface for indirect prompt injection.
- Ingestion points: Reads project configuration files including
package.json,composer.json,docker-compose.yml, and directory structures to map modules (SKILL.md, Workflow Steps 1 & 2). - Boundary markers: The skill instructions do not specify the use of delimiters or 'ignore' instructions when interpolating discovered code data into the
brownfield-map.mdreport. - Capability inventory: The agent is granted permission to read the entire source codebase and write output files specifically under the
{product_root}/docs/context/directory. - Sanitization: No explicit sanitization or validation of the names or content discovered in the codebase is performed before inclusion in the agent-dense output report.
Audit Metadata