Netraptor 0.6.0: safer recording and agent access
Version 0.6.0 · Released
Update both the Chrome extension and netraptor-mcp to 0.6.0. Restart your agent and old bridge, then list and confirm your traffic source again. Mixed versions refuse traffic reads until both components are updated.
Confirm the browser behind every read
All five MCP traffic-read tools require the browser identity from list_sessions. Pairing rotation revokes pending responses, and credential redaction covers nested JSON, forms and sensitive URL fields.
Guide: An MCP server for your web app's API traffic (Codex, Claude Code, Cursor)
Recording that cleans up after a crawl
Crawls borrow temporary recording without changing your Keep history rules. Stop and lost site access cancel queued work, while uncertain licence validation preserves saved history.
More faithful exports and generated tools
Exports preserve captured query values and destinations. Unsupported Postman or Bruno literals are reported before export. Generated tools keep distinct destinations, repeated inputs and independent path, query and body fields.
Fixes and improvements
- Saved bodies retain their truncation flag, interrupted sessions recover, and the toolbar badge counts committed captures.
- Refresh from network log recovers requests Chrome still lists; response bodies Chrome cannot provide show as unavailable.
- Existing API-derived tool name and description edits need re-entering after this upgrade; stored form-tool edits are preserved.
- Review Settings → Recording rules after upgrading and remove sites you did not intentionally enable: older crawl rules cannot be distinguished from user rules.
- To refresh the MCP client, use npx -y @branch42-labs/netraptor-mcp@latest and restart the old bridge and agent.