fish-shell-scripting
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external code, comments, and strings which may contain malicious instructions intended to influence the agent's behavior. However, it includes comprehensive mitigations.
- Ingestion points: The skill reads and processes .fish files, Fish configuration contents, and command output as source data.
- Boundary markers: The 'Instruction Authority' policy explicitly designates task material and tool output as untrusted data that cannot grant authority or expand the agent's scope.
- Capability inventory: The skill utilizes subprocess calls to shell commands such as fish, fish_indent, and the gh CLI for validation and auditing.
- Sanitization: The workflow requires using non-executing parse checks (fish --no-execute) and ensuring behavioral checks are non-mutating.
- [EXTERNAL_DOWNLOADS]: The skill describes an optional workflow for retrieving the human-facing-writing peer skill from the vendor's repository.
- Integrity: The retrieval process involves explicit user confirmation and snapshot validation (using full object IDs) to ensure content has not been tampered with.
- Trust: The resource is hosted on github.com/porada/domfiles, which is a repository owned by the skill's author.
Audit Metadata