Kiro - Introduction
Overview
Estimated time: 15–25 minutes
Kiro is a specialized, repo-aware AI coding assistant focused on developer productivity, safe automation, and team workflows. In 2025 Kiro emphasizes local augmentation (private embeddings), task templates, and gated automation so teams can scale AI assistance safely.
Core Features
Context Aware Completions
- Inline suggestions that use local repo context and recent edits
- Function-level and multi-file completions with diff previews
Task Automation
- Generate tests, refactors, and documentation via task templates
- Run sandboxed test runs and produce machine-readable results
Privacy & Local Augmentation
- Local embedding caches and private indexes for fast, private retrieval
- Workspace-level data controls to exclude sensitive files
Getting Started
- Install the Kiro extension for your editor or the Kiro CLI.
- Set your API key or workspace delegation token in secure storage (OS keychain or environment variables).
- Initialize a local code index (optional but recommended) to improve recall and reduce remote surface area.
- Enable review gates for automated tasks on protected branches.
Quick Example: Ask Kiro to add tests
Prompt:
"Add unit tests for the checkout flow covering coupon edge cases. Use pytest, mock external payment calls, and ensure tests are fast (<5s each)."
Expected behavior:
- Kiro proposes test files
- Runs tests in sandbox and returns a summary (pass/fail)
- Offers a patch you can preview and applyBest Practices
- Run Kiro in dry-run mode first for multi-file refactors and inspect generated diffs.
- Keep sensitive files out of remote processing by updating workspace exclude rules.
- Prefer small, well-scoped tasks when asking for automated refactors.