project-orientation-sweep
Pass
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a bundled shell script (
scripts/project_inventory.sh) and standard CLI tools (e.g.,git,rg,cargo,bun,go,swift) to inventory files, detect manifests, and run validation tests. These operations are restricted to the local repository environment and align with the skill's orientation purpose. - [PROMPT_INJECTION]: The skill exposes an indirect prompt injection surface as it is designed to ingest and process untrusted data from local repositories (such as manifests, source code, and log files).
- Ingestion points: Manifest files (
package.json,Cargo.toml, etc.), entrypoint source files, and repository metadata read byscripts/project_inventory.shand targeted exploration steps inSKILL.md. - Boundary markers: The skill defines repository boundaries for exploration but does not employ specific delimiters or "ignore instructions" markers when reading file contents.
- Capability inventory: The skill has the capability to execute shell commands, build tools, and test runners based on the detected project stack.
- Sanitization: There is no evidence of explicit content sanitization or instruction filtering for data read from the filesystem.
Audit Metadata