task-orchestration
Pass
Audited by Gen Agent Trust Hub on Feb 28, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No direct attempts to bypass safety filters or override system instructions were found. The skill processes task lists from .ai/*/tasks.md, which presents a potential indirect prompt injection surface as it ingests untrusted repository content. Ingestion points: .ai/[feature_name]/tasks.md. Boundary markers: None. Capability inventory: Code implementation, linting, testing, and git commits. Sanitization: None.- [DATA_EXFILTRATION]: No evidence of unauthorized data access or external transmission. Operations are restricted to the local repository and standard git workflows.- [COMMAND_EXECUTION]: While the skill orchestrates code implementation and testing, it does so through designated sub-skills; there are no direct, unsafe shell commands or arbitrary executions within the skill logic itself.- [REMOTE_CODE_EXECUTION]: No remote scripts are downloaded or executed.- [CREDENTIALS_UNSAFE]: No hardcoded API keys, tokens, or credentials detected.
Audit Metadata