rp-export

rp-export

Use this when the user wants a packaged context file for another model, review tool, or proconsult-compatible workflow.

Check workspace binding first. On mismatch, pass root=<absolute path> on every rp-mini tool call. If a shell is available and no MCP client is loaded, use node packages/server/dist/cli.js tool <workspace> <tool> --json-args '...' or its wrappers.

1. Clarify first: if export purpose, scope, preset, or budget is ambiguous, ask with AskUserQuestion before spawning the builder.
2. Spawn the context-builder subagent with:
   • task: curate export-ready context for the requested purpose
   • budget: caller budget or default discovery budget
   • response_type: question for evidence exports, plan for planning exports, or review for diff exports
   • enhancement mode: preserve when the user wants selection-only export; otherwise rewrite
3. If the builder returns <questions>, follow the question loop defined in rp-build/rp-investigate before continuing.
4. Inspect the resulting selection and token count with workspace_context include=["selection","tokens"].
5. Export with workspace_context op=export, choosing the matching preset when provided (standard, plan, review, or diff-followup).
6. Hand the payload path and receipt path to the user. Mention that the payload is compatible with proconsult-style external model intake.
7. Cite the workspace_context op=export receipt: token totals, content hash, and saved handoff profile. If the caller asked for a durable artifact, write the export to a file with the host Write tool or shell CLI.
8. Report selected scope, token count, preset, assumptions, and any files deliberately excluded.

Do not paste large exported payloads into chat; provide paths and receipt details.