defect-lifecycle-agent-skill
Pass
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implementation uses local scripts (
scripts/verify-coverage.js,scripts/export-tracker-artifact.js,scripts/validate-skill.js) that utilize standard Node.js built-in modules (fs,path). No external npm dependencies are required or installed at runtime. - [COMMAND_EXECUTION]: The skill performs command execution to run its internal validation and export scripts. These operations are scoped to the local repository environment and use files provided as command-line arguments.
- [DATA_EXPOSURE]: The skill reads repository context (code, tests, documentation) to perform its functions. There is no evidence of unauthorized access to sensitive system files (~/.ssh, ~/.aws) or exfiltration of data to external domains.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted data from repository files and user reports. It addresses this risk surface through the use of strictly defined markdown templates and JSON schemas for all output artifacts.
- Ingestion points: Untrusted data enters the agent context via repository file reads and user-provided defect descriptions as noted in
SKILL.md. - Boundary markers: The skill enforces the use of specific section headings in
assets/bug-discovery-template.mdandassets/bug-fix-report-template.mdto structure the agent's response. - Capability inventory: The skill has the capability to write files to the local repository and execute its own helper scripts via Node.js.
- Sanitization: The skill provides canonical JSON schemas in the
schemas/directory to facilitate validation of machine-readable report artifacts.
Audit Metadata