fix(provider-muse): run on Muse 1.4.6, with linked web search hits and workflow results

[t3dotgg]

Muse Code 1.4.6 shipped. T3 still ran on the 1.4.2 SDK, and a few Muse rows showed the wrong thing:

• A web search showed its raw JSON output as one result, instead of a list of linked hits.
• A finished workflow showed the label "Workflow: model-chosen generated workflow", not its result.
• A workflow agent that Muse reports as completed with a failureKind showed as completed. The MSP schema says that combination means the attempt failed.

Change

• Bump @muse-code/sdk to 1.4.6. It was published today, so it gets a minimumReleaseAgeExclude entry, like the other pinned SDKs.
• Re-record the muse_workflow replay fixture against a live Muse 1.4.6, so it carries finalSummary. The other Muse fixtures keep their 1.4.3 and 1.4.4 recordings.
• web_search output becomes one result per hit, with title, URL and snippet. web_fetch keeps one result and now links the fetched page.
• A settled workflow row shows finalSummary.
• A workflow child with terminal: completed plus failureKind shows as failed, with the failure kind as its result when Muse gives no reason.

Nothing else in 1.4.6 needs T3 changes. Its new methods (workflow/resume, task/list, recap/*, plugin/install, marketplaces, skill/import, computer-use setup) are client-called, and T3 calls none of them. A workflow can only be paused by a client, so a T3 session never sees a paused workflow. The new webSearch item kind is for provider-hosted search; a live 1.4.6 turn on Meta's provider still uses the web_search tool call. Typed quota errors are left to #18266.

Verification

• vp test run in packages/provider-muse: all pass. The 3 new tests fail on main.
• OrchestratorReplayFixtures.integration.test.ts: 140 pass. The new muse_workflow check (the row shows the recorded finalSummary) fails on main.
• All 8 Muse scenarios were recorded live against muse serve 1.4.6 through the real adapter, and all 8 passed their checks. Only muse_workflow is committed, to keep the diff reviewable.
• tsc --noEmit is clean for provider-muse and apps/server. Lint and format are clean on the changed files.
• Not checked in a real client. The clients already render web_search results and tool output; this PR only changes what the adapter puts in them.

Reviewed with sol-loop: 2 rounds with GPT-6.1-Sol on high.

Created with Claude Opus 5.5 in Claude Code.

🤖 Generated with Claude Code

[macroscopeapp]

Macroscope has since reviewed this pull request. An earlier review was skipped by a cost limit; a review has now completed, so that notice no longer applies.

[macroscopeapp]

Approvability

Verdict: Approved at 5ae14bf

Macroscope's review found this PR approvable — This is a localized Muse provider compatibility and presentation fix: it updates the SDK and correctly exposes existing search, workflow, and failure information. The remaining large changes are replay fixtures and tests, with no product-default, deployment, security, or static-analysis configuration impact.

^{You can add or adjust custom eligibility rules. Learn more.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	20.9 KiB	20.9 KiB	0 B (0.0%)	29.3 KiB	✅
Codex	Live turn messages	2	2	0 (0.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.2 KiB	21.2 KiB	0 B (0.0%)	29.3 KiB	✅
Claude	Live turn messages	2	2	0 (0.0%)	8	✅

Baseline: 2783962 · PR result: 5ae14bf · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.