Tools provided
Tools the agent exposes — your AI client calls them automatically when it needs them.
- run_full — Orchestrated path: analyse the provided CI history end-to-end — detect flaky tests, classify each root cause, build the time-limited quarantine plan, draft the fix PR — and return the full report.
- analyze — Run pure deterministic flakiness analysis over a provided CI history with no model call and return scored findings.
- detect_flaky_tests — Return only the genuinely flaky tests — those that both pass and fail with status flips — scored worst first, with regressions and stable tests excluded.
- detect_regressions — Return only the consistently failing tests that look like real regressions. These are surfaced to be fixed and are never quarantined.
- classify_root_causes — Classify each test's likely root cause — timing, shared state, network or unknown — from its failure messages, one row per test.
- build_quarantine_plan — Return the time-limited quarantine entries for flaky tests at or above the threshold: test id, root cause, flakiness rate and auto-expiry date.
- draft_fix_pr — Draft the fix PR for the quarantined set — title, branch, labels, markdown body and file changes. A draft for human review, never merged.
- list_capabilities — List the agent's static capabilities: the available tools, the root-cause classes and the default quarantine threshold.
- 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, branch, history window and quarantine threshold. The entry point for running the agent on your own data.
- run_form — Internal: invoked by the guided form when the user submits. Not for direct use.
