Tools provided
Tools the agent exposes — your AI client calls them automatically when it needs them.
- run_full — Orchestrated path: diagnose bottlenecks in the provided CI timings and return the full before/after optimization report with cache, parallelization, test-splitting and redundant-step recommendations plus a workflow-diff preview.
- analyze — Run pure deterministic bottleneck analysis over provided CI run data with no model call and return scored findings.
- build_report — Build the full deterministic before/after report: step breakdown, per-step recommendations, before and after totals, savings percentage and the workflow-diff preview.
- recommend_optimizations — Return only the per-step optimization recommendations — cache, parallelize, split-tests or remove-redundant — each with its estimated seconds saved.
- estimate_savings — Return only the before and after time proof: the current total, the estimated total after all recommendations, the seconds saved and the savings ratio.
- breakdown_steps — Return only the per-step duration breakdown — name, duration and share of the total — sorted worst first.
- preview_workflow_diff — Return only the unified-diff workflow-file edit preview for the proposed optimizations — the body of the pull request.
- list_capabilities — List the agent's static capabilities: the available tools and the bottleneck severity thresholds. Useful for discovery.
- plan_inputs — Plan the inputs for a tool: returns the clarifying questions, the JSON schema and a ready-to-edit example.
- discover_intent — Understand your goal and co-design the exact input through clarifying questions before running.
- open_form — Opens the guided form to collect the repository, the workflow file and your goal. The entry point for running the agent on your own timings.
- run_form — Internal: invoked by the guided form when the user submits. Not for direct use.
