The Agent Skills Directory

[SAFE]: The skill is authored by Zapier and utilizes the official zapier-sdk CLI tool, which is an expected and safe practice for vendor-provided skills.
[COMMAND_EXECUTION]: The skill executes shell commands using zapier-sdk and jq to fetch, test, and publish workflow configurations. These operations are restricted to the intended purpose of the skill and do not involve unauthorized system access.
[PROMPT_INJECTION]: The skill retrieves workflow source code (workflow.ts) which could theoretically contain malicious instructions (Indirect Prompt Injection). However, the skill mitigates this risk by instructing the agent to provide a detailed summary of changes and wait for explicit user confirmation before any modifications are applied or published.
[EXTERNAL_DOWNLOADS]: The skill mentions downloading diagnostic tools (e.g., workflows-doctor) from the official zapier/agent-skills GitHub repository, which is a trusted source within the vendor's own ecosystem.

workflows-modify