AI for Developers

I am building a [feature description] in a [framework] application. The relevant route or module is [path]. The repo's existing patterns are [patterns]. Before writing code, propose 2 implementation approaches with the trade-offs of each (file-count impact, test-coverage implication, future-extension flexibility). Then ask me which approach to take. Codebase context: [paste relevant files].

Diagnose this stack trace. The 3-5 most likely root causes ordered by probability, with the disambiguating test for each cause (the smallest test that rules in or out the candidate). Stack trace: [paste]. Recent code changes in the affected file: [paste]. Steps to reproduce: [paste].

Review this diff as a senior engineer at our team would review it. Explicit attention to: security (injection, auth bypass, secret exposure), performance (N+1, render-cost, bundle-impact), correctness (edge cases, error handling, race conditions), maintainability (naming, comments, abstraction-level). For each substantive issue: the line, the issue, the recommended fix, the reasoning. Diff: [paste].

I need to add [library] to this project. Before I commit to it: the 90-day maintenance signal (commits, releases, issues), the security-advisory history, the bundle-size impact, the alternative libraries with their trade-offs, the migration cost if we change our mind in 6 months. Use sourced data the developer can verify. Project context: [paste package.json or equivalent].

Generate the test suite for this function. Cover: the happy path, the 8 edge cases the function does not currently handle (empty inputs, auth boundary, race condition, partial-failure recovery, malformed data, oversize input, locale variant, accessibility regression where relevant), the error path with the specific error type, the integration-boundary cases. Use the test framework and assertion patterns the repo already uses. Function: [paste]. Existing test pattern: [paste].

I need to refactor this 400-line file into smaller, testable units without changing behavior. Walk through: the 5 responsibilities in the file with the line ranges, the proposed module split with the public interface of each module, the migration sequence that keeps the test suite green at each commit, the risks of the refactor and the mitigations. File: [paste].

Help me debug a flaky test. The test passes 70% of the time and fails 30%. Walk through: the 5 most common flaky-test causes (async race, time-of-day dependency, shared mutable state, network mock leak, fixture-ordering dependency), the diagnostic test that disambiguates each cause, the fix once the cause is identified. Test: [paste]. Test environment context: [paste].

I am deciding between [framework A] and [framework B] for [project description]. Walk through: the technical fit against the project's specific needs (data shape, scale, team size, deployment constraints), the 2026 ecosystem signal for each (community size, hiring market, library availability, long-term viability), the migration cost if we change our mind in 18 months, the recommendation with the reasoning. Project context: [paste].

Generate the architecture decision record (ADR) for this decision: [decision]. Sections: context, decision, consequences (positive, negative, neutral), alternatives considered with their trade-offs, the team and date. Voice: clear, specific, the way an ADR earns its place in the repo's decisions/ folder. Decision context: [paste].

I have a CI pipeline that takes 22 minutes. Walk through: the 5 most common slow-CI causes (test ordering, dependency install, build cache misses, integration-test sequential execution, deploy step), the diagnostic to identify the bottleneck in this pipeline, the optimization candidates with the realistic time savings of each. Pipeline config: [paste].

Generate the pull request description for this change. Sections: what changed in 2 sentences, why it changed in 1 paragraph, the testing performed (unit, integration, manual, screenshots), the rollback plan if the change breaks production, the reviewer checklist for security, performance, and correctness. Voice: clear, scannable, the way the PR description lands as the merge-commit message. Diff and context: [paste].

I am evaluating whether to take on [side project] alongside my day job. Walk through: the realistic time investment for the first 90 days, the technical-skill-development value against my current trajectory, the portfolio-and-network value, the legal and IP considerations against my employment agreement, the recommendation with the reasoning. Project and context: [paste].