Engineering templates

Assign backend or API work and the engineer maps the existing architecture, tests the public contract, implements the smallest safe change, runs the project checks, and drafts a pull request for review.

Assign a bug and a coding agent investigates the root cause, proves it with a failing test, fixes the smallest surface, self-reviews the diff, and drafts a pull request for you to merge.

Assign a schema or data migration and the engineer plans the change, writes the migration in the repo's toolchain, verifies ordering and rollback risk, updates affected code, and drafts a PR.

On the schedule you choose, the agent diffs shipped changes against the docs, drafts the reconciling updates and API endpoint docs typed to the real response, edits for clarity, and drafts a PR — so docs never silently drift from the code.

Assign a frontend task and the agent builds or optimizes the feature with composable, accessible, performant components, self-reviews, and drafts a deploy-ready PR for you to review — it never deploys on its own.

On the schedule you choose, the agent classifies and dedupes incoming issues, verifies the actionable ones, proposes labels and routing under explicit rules, and drafts a triage summary for you to approve before anything is applied.

Assign model work and the engineer builds a reproducible pipeline with baselines, leakage checks, metric-driven evaluation, error analysis, and a review-ready PR or report.

Assign mobile app work and the engineer implements it with platform-native UX, safe areas, performant lists and gestures, offline/network behavior, tests, and a review-ready PR.

When a pull request opens, a reviewer reads the diff from your repository, runs a correctness, security, and standards pass, and drafts specific inline comments plus a verdict for you to approve before anything is posted.

Assign a feature, release, or PR and the QA reviewer builds a risk-based test plan, exploratory charters, edge-case matrix, and coverage-gap report for approval.

Point the agent at a messy area and it reads your code, finds the shallowest module, writes a tiny-commit refactor plan, applies it behind your tests step by step, and drafts a PR — so debt comes down without behaviour changing.

Assign a branch, PR, feature area, or repo slice and the reviewer traces attacker input to sinks, checks auth and CI risks, scans dangerous patterns where possible, and returns high-confidence findings with fixes.

Hand the agent a spec and it turns each acceptance criterion into a failing test, watches it fail, writes the minimal code to pass, refactors green, and drafts a PR — features shipped test-first.

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.