Tools provided
Tools the agent exposes — your AI client calls them automatically when it needs them.
- classify_failure — Read a failing CI run's log and say whether the failure belongs to the test, the code under test, or the infrastructure, citing the exact log line that decided it — or return an explicit unclassified verdict.
- detect_flaky — Find the unstable tests in a CI run history: measure each test's flip rate, classify its likely root cause, separate a real regression from a flake, and propose a time-limited quarantine.
- reproduce_bug — Turn a bug report into a minimal reproduction: ordered repro steps, the affected code path, a failing test that proves the bug before any fix, and a confidence score.
- find_duplicates — Score a new bug report against the pool of known issues you supply, ranking each by term-frequency similarity and naming the terms they share.
- verify_fix — Judge independently whether a bug actually closed: compare the original failure evidence to the applied diff, and flag a fix that merely deletes the assertion that caught the bug.
- remember_incident — Apply the prevention rules implied by your closed incidents to one change: scan the added lines of a unified diff and cite every line that reproduces an incident's snippet, symbol or file.
- find_coverage_gaps — Rank the untested code paths in a coverage report by the business risk they carry, weighting each file's gap by criticality and by how many lines are actually at stake.
- audit_mutation — Audit the mutation-testing results you supply: name the tests that covered a surviving mutant and passed anyway, and score detection over the mutants that actually counted.
- generate_test_cases — Derive positive, negative, boundary and edge test cases from one requirement, producing one traceable case per equivalence class, each citing the clause it came from.
- write_tests — Write test source from one of three kinds of evidence you choose: the code under test, a bug report as a single failing test, or coverage gaps. Nothing is written to disk or executed.
- parse_artifact — Parse a pasted build artifact — an LCOV coverage report, an npm audit JSON report, or a Terraform plan JSON — into structured records, reporting how many rows were understood.
- run_full — Run every capability you supplied the input for, in registry order, and return one report. Capabilities whose input is missing are listed as skipped.
- list_capabilities — List what this agent can do: every tool it advertises, what each one is for, and the arguments each one reads.
- plan_inputs — Plan the inputs for a tool: returns the questions to ask, the JSON schema of the arguments, and a ready-to-edit example.
- discover_intent — Understand your goal and co-design the exact input through clarifying questions before anything runs.
- open_form — Opens an interactive form that collects the run arguments. The form is the entry point for running the agent.
- run_form — Internal: invoked by the open_form form when the user submits. Not for direct use.
