fix: MCP 工具错误提示不清晰导致无法定位参数问题

[binggg]

Attribution issue

• issueId: issue_mo8whjxm_vaf900
• category: tool
• canonicalTitle: MCP 工具错误提示不清晰导致无法定位参数问题
• representativeRun: atomic-js-clientsdk-init-login-create-collection/2026-04-21T17-30-30-b3vf6z

Automation summary

• root_cause: The buildCapiErrorMessage function in mcp/src/tools/capi.ts didn't handle the generic "400 invalid parameter value" error pattern. When the Tencent Cloud API returned this error (e.g., "400 invalid parameter value (request-id)"), the MCP tool just passed it through without providing helpful context about which parameter might be wrong or what to do next, making it impossible for agents to locate the parameter problem.
• changes:

1. Added detection for the "400 invalid parameter" error pattern using regex /400\s+invalid\s+parameter/i
2. Added helpful guidance for this error pattern that includes:

• Explanation that the parameter value is invalid or doesn't meet API requirements
• A checklist of things to check (parameter type, format, enum values, required fields)
• Parameter hints for known TCB actions (shows common and required parameter keys)
• Suggestion to use searchKnowledgeBase(mode="openapi", query="${action}") to find detailed API documentation

3. Added test cases in mcp/src/tools/capi.test.ts to verify the new error handling

• validation:
• All 16 tests pass (including 4 new test cases for the 400 error handling)
• TypeScript compilation passes with

Changed files

• mcp/src/tools/capi.test.ts
• mcp/src/tools/capi.ts

[binggg]

Attribution post-PR evaluation

• visibility: internal identifiers, run ids, and private links are intentionally omitted
• attempt: 1
• eval_scope: primary_only
• overall: FAILED
• summary: at least one planned evaluation case failed
• updated_at: 2026-04-28T23:35:08.299Z

Cases

• [FAILED] — primary — evaluation failed

[binggg]

Closing stale automated attribution PR. The issue has been superseded by subsequent work or is no longer actionable.