Test gap filler
On the schedule you choose, the agent finds the untested behaviours that matter, backfills coverage one slice at a time in your conventions, self-reviews, and drafts a PR — so coverage keeps rising without a human writing every test.
Saves you ~4.6 h / week
What it installs
Agents 2
-
Coding Agent
Backfills behaviour-level test coverage and drafts a pull request for approval.
-
Test gap filler Quality Reviewer
Checks primary evidence, domain controls, deliverable completeness, and communication quality, stopping the run when the work is wrong, unsupported, incomplete, or uncertain.
Teams 1
-
Test gap filler quality team
The delivery agents produce the work while an independent quality reviewer checks each workflow handoff against explicit evidence, domain, and communication requirements before the run can continue.
Workflows 1
-
Fill test gaps
Find the behaviours that matter, backfill them in vertical slices, self-review, and draft a PR.
Goals 1
-
Coverage rising
Coverage of critical-path behaviour increases over time, with a green suite.
Skills 3
-
test-scenarios
Turn each behaviour into an explicit test scenario — objective, starting conditions, role, step-by-step actions, observable outcomes, and edge cases — before writing the assertion. Adapted from phuryn/pm-skills/test-scenarios.
-
tdd
Test-first discipline applied to backfilling coverage: vertical slices over horizontal, behaviour over implementation, one red-green cycle at a time, refactor only when green. Adapted from mattpocock/skills/tdd.
-
qa
Quality-assurance reading of the codebase to find what is genuinely untested and user-facing, framed as durable behaviour rather than implementation. Adapted from mattpocock/skills/qa.
Schedules 1
-
Fill test gaps
Finds untested behaviours and drafts coverage on the schedule you choose. ## Professional quality control Objective: complete the recurring Test gap filler operating cycle and deliver a review-ready, evidence-backed result rather than merely report that the schedule ran. Start by fixing the review period, reading the current source records and prior run, confirming required access, and listing missing or contradictory inputs. Execute the bundle's full authored method in order, retaining source, calculation, command, or before/after evidence for load-bearing findings. Separate facts from interpretation, apply supplied policy without inventing thresholds, and stop any branch that requires missing authority or an unconfirmed human rule. Produce the complete named deliverable plus an evidence record, assumptions, exceptions, unresolved questions, and concrete next actions with owners where the method calls for them. Run the workflow's independent verifier checks and fix failures before requesting approval. The task is done only when every requested section and quality criterion passes or a blocker is explicit; silence, inaccessible data, or polished prose is not proof of completion. Inspect the repository and actual diff, reproduce the behavior, run the declared tests and checks, cover known success and error paths, and verify security, performance, migration, and rollback implications where relevant. Never infer a passing command. Technical prose must cite command output and avoid generic AI phrasing or claims that the code does not prove. The human retains authority over publication, sending, spend, signing, merging, deployment, and other consequential external actions.
Requirements
What this playbook expects to do its job. Task Machine does not verify these — you decide whether your setup is ready.
- Connected repository — Needs a connected code repository so the agent can read the source, the existing test suite, and run the project's checks while it backfills coverage.
Setup guide
How to Fill Test Gaps With an Agent
A practical guide to finding untested behavior, writing scenarios, backfilling coverage, and drafting reviewable PRs.
Read the setup guideDon't see your workflow? Describe it.
A sentence or two about a recurring job is enough. We design the playbook that runs it and show you exactly what it saves.
Install Test gap filler and run it with approvals.
Join the waitlist and we will send early access when the first private beta spots open.
Private beta. We invite teams in batches and never share your email.