ziw-setup
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill incorporates defensive instructions via an 'Instruction Trust Boundaries' policy, requiring the agent to disregard policy override attempts found in untrusted contexts like issue comments or CI logs.
- [DATA_EXFILTRATION]: Implements a 'Safety' policy that strictly prohibits the inclusion of secrets, tokens, signed URLs, or private customer data in the generated configuration files.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from external systems.
- Ingestion points: Retrieves content from issue trackers (Linear, GitHub), PR comments, and CI log outputs.
- Boundary markers: Present and clearly defined in SKILL.md and references/operating-profile.md, establishing a hierarchy of trust.
- Capability inventory: Authorized to write to docs/agents/workflow/config.md and execute read/write tool calls against issue trackers.
- Sanitization: Instructs the agent to ignore malicious instructions embedded in data and report them as friction findings.
Audit Metadata