feat(btw): assign MCP server tools to conversation or work

[YUZHEthefool]

Preflight

• Existing Issues and PRs were searched.
• This follows the current-branch, no-legacy, Python 3.14+ policy.
• Exactly one Agent note is provided by the actual author.

Problem

Tools from a single MCP server should have an explicit, consistent loop assignment rather than appear in conversation merely because the server is installed.

Proposal

Assign MCP capabilities by server to conversation, work, or both, with work-only as the proposed default. Deliver the profile selector and actual enforcement as one slice.

Design: PR #28 and its English proposal. Parent: P / #122. PR #28 is documentation only; this issue tracks follow-up implementation or evaluation.

Associated PR: PR #144 (non-draft; base split/btw-plugin-routes).

Acceptance criteria

• All tools belonging to a selected MCP server follow the same saved assignment in both main-agent and handoff execution.
• Default work and explicit both round-trip through profile save/reload; disabled or unavailable servers have a defined UI and runtime outcome.
• Disabling BTW retains current MCP behavior. Loop assignment does not change MCP read/write authorization or URL, DNS, redirect, and private-network controls.
• Use the current tool catalog and existing server-list API; UI tests and bilingual hints describe the same behavior.

Dependencies and delivery

Blocked by: B1 / #123, B7 / #129

Coordinate the shared capability selector with B10 / #132 while keeping server assignment separately reviewable.

Implementation anchors

Historical prototype 33ee103a62937db3e930c89ba47a648b75cc7772 provides provenance, not code to copy unchanged:

• astrbot/core/astr_main_agent.py: prototype _filter_mcp_tools_for_loop.
• astrbot/core/astr_agent_tool_exec.py: handoff MCP filtering.
• dashboard/src/components/shared/CapabilityLoopSelector.vue: server_name assignment.

Integrate with the current tool catalog, Skill snapshot assembly, and authorization service, rechecking current master when implementation begins.

Validation

Commands below are required for the implementation PR, not results claimed by this issue.

Existing current-master regression baseline:

uv run pytest tests/unit/test_astr_main_agent.py tests/unit/test_astr_agent_tool_exec.py tests/unit/test_skill_tool_assembly.py

Historical prototype tests to adapt with the feature (they are not shipped by documentation-only PR #28):

• dashboard/tests/capabilityLoopSelector.vitest.ts

After adapting the corresponding Dashboard test, run from dashboard/:

pnpm test:vitest tests/capabilityLoopSelector.vitest.ts
pnpm i18n:check

For changed documentation, run pnpm run docs:build from docs/ and make check-md from the repository root. Update OpenAPI and generated outputs together only if the HTTP contract changes.

Out of scope

Per-tool policy redesign, additional MCP transports, and network-policy changes are outside scope.

Repository requirements

Follow AGENTS.md, GOVERNANCE.md, and AI_POLICY.md. This issue uses the feature_request template. Keep Python 3.14+, current configuration shapes, existing authorization boundaries, and bilingual feature documentation. A follow-up PR requires human maintainer review and a separate AI-assisted review. Use the feat PR template for the corresponding work.

Author

Agent

Agent note

Codex prepared this development issue at the user's request from PR #28, its discussion, the referenced source, existing tests, and current-master integration points. AI assisted the research and English drafting. Filing this issue changes no runtime behavior and does not claim feature tests or model trials have passed. Remaining decisions concern implementation details and experimental evidence within the stated scope.