fix(test): mount pending MCP App calls

[ScriptedAlchemy]

Closes #736.

Summary

• Make MountBrowserAppOptions.toolResult optional: omission mounts the opening call in flight, publishes its input, and leaves cancellation/result/error to the mounted harness publishers.
• Derive canonical hostContext.toolInfo.tool from toolName / toolDefinition, fill the existing object-schema default for partial definitions, merge caller host context, and reject conflicting tool names.
• Name invalid MCP App host-context fields in bridge construction errors instead of reporting only generic unstable values.
• Prove cancellation through the convention-first MCP App, preserve initial-result behavior, and document the public contract in English and Chinese.

Proof

The browser-App suite observes onToolInput render checking, then proves each terminal path independently: cancellation renders through onToolCancelled, a later result renders degraded, and a later isError result renders unavailable. The existing initial result still arrives during mount and rejects later cancellation. It also verifies derived full tool metadata, caller-context merging, partial tool definitions, and conflicting metadata rejection.

Local merge gate

Branch contains origin/main at 240d7b0a97.

pnpm build && pnpm typecheck && pnpm lint && pnpm test:unit
# passed: build, typecheck, lint (1,493 files), 4,482 tests / 0 failed (4,476 passed, 6 skipped)

AGENT_BUNDLE_WORKBENCH_PREBUILT=1 AGENT_BUNDLE_PACKAGE_PREBUILT=1 pnpm exec rstest --config rstest.integration.config.ts \
  packages/agent-bundle/tests/examples-contract.test.ts \
  packages/agent-bundle/tests/mcp-apps-compile.test.ts \
  packages/agent-bundle/tests/serve-app.test.ts \
  packages/workbench/tests/mcp-app-preview-browser.test.ts \
  packages/workbench/tests/mcp-app-frame.test.ts \
  packages/workbench/tests/mcp-app-real.e2e.test.ts \
  packages/workbench/tests/examples-real.e2e.test.ts \
  packages/workbench/tests/web-command.e2e.test.ts
# passed: 60 / 60

pnpm --filter @agent-bundle-example/mcp-app check
pnpm --dir examples/mcp-app test:browser-app
# passed: validate/build/typecheck; 14 / 14 browser-App tests

pnpm docs:site:build
# passed: locale parity, 0 broken links / 29,989 anchors

Deslop

Deslop: GPT-5.6 Sol, 1 edit (moved the opening-result test helper before its first use).

Self-review

Reviewer: Claude Fable 5.1 Thinking (high), change-risk-reviewer, two passes.

Finding	Disposition
Medium: deriving toolInfo made a name-only toolDefinition fail for missing inputSchema	Fixed by filling the harness's existing { type: 'object' } default; browser regression added.
Low: example prose made opening-tool gating appear to apply to onToolCancelled	Fixed in English and Chinese; route listeners and global cancellation are now distinct.
Low: caller toolInfo conflict detection is name-only	Accepted: the opening tool name is the routing identity; the canonical definition always comes from harness options, and docs tell callers not to duplicate it.
Low: explicit deferred terminal state is not replayed after a second ui/initialize	Accepted: mountBrowserApp models one mounted opening call; no rebind surface is exposed by the harness.
Info: pending bindings carry an internal empty result for loadResource() fallback	Accepted: browser-App mounts compiled srcdoc and never resolves its resource through that fallback; no placeholder reaches the App wire.
Low: default display context now survives a partial caller context	Accepted and documented: merging defaults plus caller fields is the requested harness behavior.

Second pass: both fixes verified; no remaining merge blockers.

[changeset-bot]

🦋 Changeset detected

Latest commit: 398c99b

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 2 packages

Name	Type
agent-bundle	Patch
create-agent-bundle	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-07T09:03:09.825828Z	398c99b	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 398c99b16a

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".