- Enrich description with trigger keywords for better auto-activation
- Add user-invocable: false (background knowledge, not a user command)
- Add "When to Use" section with concrete trigger scenarios
- Add "Constructing an Effective Search Query" with good/bad query examples
- Add "Response Structure" documenting actual search/get API return formats
- Simplify Selector Priority to a concise ordered list (no CLI-specific code)
- Generalize Fallback Strategy to use agent's own browser tools
- Add Login Page Handling section
- Add end-to-end Example using plugin tool calls
- Update system guidance in plugin.ts to be self-contained
- Update tests to match new SKILL.md content
URL.hostname wraps IPv6 addresses in brackets (e.g. [fd00::1]),
which bypassed the regex-based private IP check. Strip brackets
before matching.
Addresses: PR #213 Codex review comment
- Add private IP / localhost / non-http protocol validation in resolveApiUrl
- Unify error handling via formatToolError using isActionbookError from SDK
- Wrap register() in try-catch to prevent crashing the host on bad config
- Add 28 new tests covering tool execution, SSRF, proxy bypass, and error paths
github-actions[bot]