skills.rest
scapia-oss
scapia-ossOFFICIAL
Explore all skills by this author →
WebsiteGitHub

kiro-impl

Implements approved spec tasks using TDD with subagent dispatch and verification gates.

Updated Aug 27, 2026
Search Tags:#tdd#spec-driven#code generation#implementation#verification#task execution#subagent
One-click install
npx skills add https://github.com/scapia-oss/compass --skill kiro-impl-scapia-oss
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: kiro-impl
Source: https://github.com/scapia-oss/compass/tree/main/tools/cc-sdd/templates/agents/claude-code-skills/skills/kiro-impl
Command: npx skills add https://github.com/scapia-oss/compass --skill kiro-impl-scapia-oss

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Turning an approved spec into working code often loses discipline: tests get skipped, tasks drift from requirements, and slow build stacks make full test suites run per task. This Skill executes approved tasks.md work items with strict TDD (RED/GREEN), milestone-level batching, and evidence gates so implementation stays aligned with requirements and design. ## Core Features & Use Cases - Autonomous milestone execution: Dispatches fresh implementer subagents per milestone with a parent-witnessed RED phase, scoped GREEN gates, and one full build/test run at the end of the run. - Manual and review modes: Run selected task numbers directly in the main context, and opt into per-task review with --review inline or --review required (review is off by default). - Commit policy control: Honors the spec's commit_policy (per-task or leave-uncommitted) with --commit/--no-commit per-run overrides, and never commits to protected branches. - Use Case: After approving tasks for a feature spec, run the skill to implement all pending milestones autonomously — it writes tests first, implements against them, verifies completion, updates tasks.md markers, and commits per milestone. ## Quick Start Ask the AI to implement all pending tasks for your feature spec, for example by saying: run kiro-impl for the user-notifications feature with review inline.

Frequently Asked Questions about kiro-impl

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I implement approved spec tasks with TDD automatically?▼

Invoke the skill with a feature name and no task numbers to run autonomous mode. It dispatches an implementer subagent per milestone that writes all tests first (RED), implements against them (GREEN), and passes a scoped verification gate before the parent commits.

How do I run only specific tasks instead of the whole spec?▼

Pass task numbers after the feature name, such as 1.1 or 1,2,3, to enter manual mode. Selected tasks execute directly in the main context instead of through subagent dispatch.

Does it run a full build and test suite for every task?▼

No. Each milestone runs a scoped GREEN gate covering only its own tests, boundary test targets, and affected-module build. The full build and full test suite run once at the end of the run, which keeps slow stacks like Gradle fast.

Can I stop it from creating git commits?▼

Yes. Set commit_policy to leave-uncommitted in spec.json, or pass --no-commit for a single run. The skill implements, gates, and updates tasks.md markers but leaves all changes in the working tree for you to stage and commit.

Why is per-task review not running by default?▼

Review is opt-in and defaults to off; the standing quality gates are kiro-verify-completion and /kiro-validate-impl. Enable a reviewer with --review inline (main context) or --review required (fresh reviewer subagent per task) for risky changes.

What happens if the test suite already fails before implementation starts?▼

The skill runs a scoped pre-flight baseline before writing code. Pre-existing failures are recorded as BASELINE_FAILURES so later gates only count new failures as regressions; broad or build-blocking failures halt the run with an upstream-breakage report.

Related Skills

Explore More Software Engineering Skills→
ajsai47ajsai47
11

brainstorming

Guide collaborative dialogue to refine ideas into approved design specs.

Community
Advanced
B0yZ4kr14B0yZ4kr14

speckit-retrospective-analyze

Generates retrospective analyses and grade reports for submitted assignments and exams.

Community
Advanced
SirTheOracleSirTheOracle

proposal-reviewer

Audit final plans against the codebase for correctness and regressions.

Official
Advanced
bingeli1379bingeli1379
2

quick

Execute small-to-medium tasks with inline orchestrator analysis and minimal spec artifacts.

Community
Intermediate