Run details

aiWorkspace

CI Pipeline Executions
- Current executionSucceeded
  feat/harness-p14-mcp-server
- Runs and Agents
- TimelineBeta
- Configuration
- Self-Healing CI
- Project Graph
- Circular DependenciesEnterprise
- Resource Usage
- Analysis
- ConformanceEnterprise
Runs

Insights

Compare tasks
Analytics

fix(ai-mcp): concrete .d.ts import path, and cover the harness MCP server paths The .d.ts of createHarnessMcpServer inlines the createMCPServer return type. That type names MCPHandleOptions, and the emit picked the barrel as './server.js', which does not resolve. Import createMCPServer from './server/index', so the emit writes './server/create-server.js'. Add tests for the harness MCP server: reject, approve with no approval, resolve with an unknown id or a harness refusal, answer refusals and a question that no call waits on, elicitation "no" and decline, steer, cancel, status with a running agent and a queued turn, errors from commands, agents, and the model, a harness with no exposed agents, name sanitizing, the server name and version, and the default thread. The branches that stay uncovered in harness.ts are guards that the SDK input check and defineHarness make unreachable. To keep the ai-mcp coverage gate, also cover untested edges in pool (prefix '' and a duplicate tool name), resources (a blob without a uri), prompts (a message without content), and isMCPInputRequiredError (no kind). Remove the casts and the computed expected values in the resources and prompts tests. Claude-Session: https://claude.ai/code/session_01APYv1qshKyjPPpkFyRZhfZ

nx run-many --targets=build --exclude=examples/**,testing/**

Succeeded

4 CPU cores

read-write access token used

dceed99c1554

97%

Cache hits

71 of 73 tasks used cache.

Get faster results

Learn how to enable distribution.

Atomizer enabled

73 groups of tasks optimized.

© 2026 - Nx Cloud

Terms of Service Privacy Policy Changelog Status Docs Contact Nx Cloud Pricing Company @NxDevTools

aiWorkspace