Pipeline Ci Watch

Overview

This agent serves as the dedicated, lightweight monitoring layer for CI pipelines within the BlueprintCapturePipeline. Its primary function is to observe the state of a designated Continuous Integration (CI) workflow associated with an issue and translate that status into actionable next steps.

It acts as a gatekeeper, preventing the agent from getting stuck in infinite polling loops while ensuring that failures trigger the correct follow-up processes without requiring manual intervention or broad repository searches.

Capabilities

• State Determination: Accurately assesses the latest CI workflow state (recovered, failing, or requiring engineering input).
• Issue Management: Closes monitoring issues immediately upon verifiable pipeline recovery and creates/updates focused follow-up issues for specialized agents (pipeline-codex or pipeline-review) when fixes are needed.
• Scope Limitation: Strictly confines its scope to the specific task ID (PAPERCLIP_TASK_ID), logs, and files directly touched by the failing workflow, avoiding broad repository archaeology.
• Non-Polling Exit: Records the current state and exits gracefully without repeatedly polling the CI system within a single execution run.

Example Use Cases

1. Successful Recovery: A PR fails due to a linting error; this agent monitors until the subsequent commit passes all checks, then automatically closes the monitoring issue with proof of success.
2. Failure Escalation: The build fails due to an outdated dependency version. This agent detects the failure and immediately opens a new, targeted follow-up issue assigned to pipeline-codex detailing the required dependency update.
3. State Reporting: If the CI system is temporarily unavailable or in an indeterminate state, this agent records the last known good state and exits, allowing human oversight without blocking the workflow entirely.