Jenkins pipelines with manual gates are common and they break most integrations: the build is neither running nor finished, and a poller reports it as in progress forever. Finding pending actions and submitting an approval turns that from a dead end into a step the agent can complete or escalate. Line-level coverage is the other detail worth having — a coverage percentage tells you there is a problem and covered lines for a specific file tells you where. Tool descriptions are in Spanish.
A Jenkins client covering jobs and builds, pipeline monitoring including input steps, and code coverage retrieval at three levels of detail.
- Jobs started against a specific branch, stopped, and inspected — with jenkins_get_git_branches listing what is available to build.
- jenkins_wait_for_build blocks until a build finishes or detects a pause, which is what makes a build step in an agent workflow synchronous.
- Input steps handled properly: jenkins_get_pending_actions finds a build waiting on approval, and jenkins_submit_input_action approves or rejects it.
- jenkins_pipeline_monitor covers stages, nodes and pending inputs together, and per-build tools read step and node status individually.
- Coverage at three granularities — the report, the file paths that have coverage, and the covered lines within a specific file.
- jenkins_job_manager and jenkins_build_manager consolidate the list, read, create and update operations rather than spreading them across many tools.
A reachable Jenkins instance with credentials for it.
One command plus a key — npm install -g @grec0/mcp-jenkins, then supply credentials