Big News: Socket raises $60M Series C at a $1B valuation to secure software supply chains for AI-driven development.Announcement
Sign In

kspec

Package Overview
Dependencies
Maintainers
1
Versions
61
Alerts
File Explorer

Advanced tools

Socket logo

Install Socket

Detect and block malicious and high-risk dependencies

Install

kspec

Spec-driven development workflow for Kiro CLI

latest
Source
npmnpm
Version
2.2.4
Version published
Weekly downloads
91
-18.75%
Maintainers
1
Weekly downloads
 
Created
Source

kspec — Spec-Driven Development for Kiro CLI

npm version License: MIT

Spec-driven development workflow for Kiro CLI with context management, verification at every step, and Jira integration.

What's new in 2.2.0

AreaWhat you get
Enterprise governancekspec init --enterprise (or KSPEC_ENTERPRISE=1) — single opt-in prompt that scaffolds an MCP-registry / model-registry / IdP / prompt-logging governance steering doc auto-loaded into every agent prompt
CI/headless modekspec init --ci scaffolds .github/workflows/kspec-review.yml running kspec review --no-interactive on every PR + a CI hooks preset (preToolUse audit + destructive-command block)
Least-privilege agentsEvery agent now ships with toolsSettings: write paths scoped per-role, shell commands allow/denylisted, secrets/git/node_modules always-denied, explicit subagent.availableAgents delegation graph
Kiro Agent Skills5 SKILL.md files (/kspec-spec, /kspec-build, /kspec-review, /kspec-verify, /kspec-jira) — auto-become slash commands in CLI 2.1+ default chat
All-MCP injectionkspec-spec and the spec/build/review pipeline now see every configured MCP (not just Atlassian) — @github, @confluence, @slack, etc.
kspec sync-agentsRefresh agent JSON + IDE markdown after adding a new MCP — no full kspec init re-run
IDE chat subagentsOptional .md agent files alongside .json so Kiro IDE chat can use kspec workflows
Smart steeringinclusion: fileMatch so api-standards.md, frontend.md, backend.md only auto-load when matching files are touched. Existing inclusion_mode files migrated transparently

See CHANGELOG for the full release notes.

Why kspec?

AI coding assistants forget context, drift from requirements, and repeat mistakes. kspec solves this:

Problemkspec Solution
Context lossCONTEXT.md survives AI context compression
Scope creepSpecs define boundaries before coding
No verificationVerify at every step (spec → design → tasks → build)
Lost learningsmemory.md compounds knowledge across projects
Enterprise silosJira integration bridges BA/PM and developers

Read the full Methodology or see a complete Example Walkthrough.

Installation

npm install -g kspec

Quick Start

kspec init                    # Interactive setup
kspec analyse                 # Analyse codebase
kspec spec "User Auth API"    # Create specification
kspec design                  # Create technical design (optional)
kspec tasks                   # Generate tasks
kspec build                   # Execute with TDD
kspec verify                  # Verify implementation
kspec done                    # Complete & harvest memory

Two Ways to Use kspec

1. CLI Mode (Outside kiro-cli)

Run kspec commands from your terminal:

kspec init
kspec spec "User Authentication"
kspec design                       # optional — creates design.md
kspec tasks
kspec build

Stay inside your kiro-cli session and switch between specialized agents. Every agent includes pipeline navigation suggesting the next step:

$ kiro-cli

> /agent swap kspec-spec
> Build a todo app with categories
  (agent creates spec.md, spec-lite.md, updates context)
  → Next: /agent swap kspec-design or /agent swap kspec-tasks

> /agent swap kspec-design
  (reads spec → creates design.md with architecture)
  → Next: /agent swap kspec-tasks

> /agent swap kspec-tasks
  (reads CONTEXT.md + design.md → generates tasks)
  → Next: /agent swap kspec-build

> /agent swap kspec-build
  (reads CONTEXT.md → continues from current task)
  → Next: /agent swap kspec-verify

This approach solves the context loss problem — agents read .kiro/CONTEXT.md automatically to restore state after context compression. You never need to exit kiro-cli; the full pipeline is available through agent swapping.

Workflow

init → analyse → spec → verify-spec → design (optional) → tasks → verify-tasks → build → verify → done

Commands

Core Workflow

CommandDescription
kspec initInteractive setup (date format, execution mode, Jira project, enterprise opt-in)
kspec init --enterpriseSkip the prompt — go straight into governance setup (MCP/model registries, IdP, prompt logging)
kspec init --ciSetup with GitHub Actions workflow + CI hooks preset (audit + destructive-block)
kspec sync-agentsRefresh agent JSON/markdown after adding a new MCP server (idempotent)
kspec analyseAnalyse codebase, update steering docs
kspec spec "Name"Create spec.md + spec-lite.md
kspec verify-specInteractively review and shape spec with clarifying questions
kspec designCreate technical design from spec (optional)
kspec verify-designVerify design against spec requirements
kspec tasksGenerate tasks.md from spec (uses design.md if present)
kspec verify-tasksVerify tasks cover spec
kspec buildExecute tasks with TDD
kspec verifyVerify implementation matches spec
kspec doneComplete spec, harvest memory

Jira Integration (requires Atlassian MCP)

CommandDescription
kspec spec --jira PROJ-123,PROJ-456 "Feature"Create spec from Jira issues
kspec sync-jiraSmart sync — updates existing issue or creates new
kspec sync-jira --createForce create new Jira issue
kspec sync-jira --project SECOPSCreate in specific project
kspec sync-jira --update PROJ-123Update existing Jira issue
kspec jira-pullPull latest updates from linked Jira issues
kspec jira-subtasksCreate Jira subtasks from tasks.md
kspec jira-subtasks PROJ-123Create subtasks under specific issue

Work Types (Abbreviated Pipelines)

CommandDescription
kspec fix "Bug description"Bug fix with TDD (spec→test→fix→verify)
kspec refactor "What and why"Refactor code (no behavior change)
kspec spike "Question"Time-boxed investigation (no code)
kspec reviseRevise spec from stakeholder feedback
kspec demoGenerate stakeholder walkthrough
kspec estimateAssess complexity before building

Memory, Milestones & Observability

CommandDescription
kspec memoryShow project memory
kspec memory reviewAI-assisted memory review
kspec memory pruneRemove outdated entries
kspec milestone listList milestones
kspec milestone create <name>Create milestone
kspec milestone add <name>Add current spec to milestone
kspec milestone status <name>Show milestone progress
kspec metricsShow timeline for current spec

Agentic Review Loop

CommandDescription
kspec review [target]Code review with agentic loop (if reviewers configured)
kspec review --simpleQuick review without loop
kspec analyseAnalyse codebase with review loop
kspec analyse --no-reviewSkip review loop
kspec build --reviewBuild with agentic review loop

Other

CommandDescription
kspec refreshRegenerate spec-lite.md after editing spec.md
kspec contextView/refresh context file
kspec listList all specs
kspec statusPipeline-aware status with next step suggestion
kspec agentsList available agents
kspec updateCheck for updates
kspec helpShow help

Work Types

Not everything needs the full spec pipeline. kspec provides entry points for different work types:

Bug Fix

kspec fix "Login fails with special characters"

Or in kiro-cli:

> /agent swap kspec-fix
> Login fails when email contains + character. Error in auth.js line 42.

Refactor

kspec refactor "Extract validation logic from controllers"

Or in kiro-cli:

> /agent swap kspec-refactor
> Extract all validation logic from route controllers into a shared validation middleware

Spike (Investigation)

kspec spike "Can we migrate from REST to GraphQL?"

Or in kiro-cli:

> /agent swap kspec-spike
> Investigate feasibility of migrating our REST API to GraphQL

Revise (Feedback Loop)

kspec revise    # revise current spec from feedback

Or in kiro-cli:

> /agent swap kspec-revise
> PM says we need to add OAuth support and remove the remember-me feature

Demo (Stakeholder Walkthrough)

kspec demo      # generate walkthrough of current implementation

Estimate (Complexity Assessment)

kspec estimate  # assess before committing to build

Design Pipeline

The optional design step sits between spec and tasks, enabling technical architecture planning before implementation:

kspec spec "Payment Processing"    # Create spec
kspec design                       # Create design.md (architecture, data models, APIs)
kspec verify-design                # Verify design covers spec
kspec tasks                        # Generate tasks (uses design.md for ordering)

design.md includes:

  • Architecture Overview
  • Component Breakdown
  • Data Models
  • API Contracts
  • Dependency Mapping
  • Technical Decisions
  • Risk Assessment

The design step is optional — run kspec tasks directly to skip it. When design.md exists, the tasks agent uses it for architecture guidance and dependency ordering.

Interactive Spec Shaping

kspec verify-spec goes beyond simple PASS/FAIL verification. It interactively shapes your spec:

  • Reads your spec.md thoroughly
  • Asks 4-8 targeted clarifying questions with sensible defaults
  • Proposes assumptions: "I assume X, is that correct?"
  • Waits for your responses
  • Suggests specific updates to spec.md
  • Gets your confirmation before making changes

This ensures specs are complete and unambiguous before moving to design or tasks.

Jira Pull Updates

Keep specs in sync with evolving Jira requirements:

kspec jira-pull

This fetches the latest state of all linked Jira issues, generates a change report showing new/modified criteria, status changes, and comments, then presents changes for your approval before modifying spec.md. Specs are never auto-updated.

Agentic Review Loop

kspec implements a devil's advocate pattern for code review, using multiple AI CLI tools as reviewers.

How It Works

  • Doer (kspec-review agent) performs the initial review
  • Reviewer (configured external CLIs) critiques the doer's work
  • Loop continues up to 3 rounds until approved or questions remain
  • Human-in-the-loop surfaces unresolved questions for your input

Configure Reviewers

During kspec init, select which CLIs to use as reviewers:

  • GitHub Copilot CLI (copilot)
  • Gemini CLI (gemini)
  • Claude Code CLI (claude)
  • OpenAI Codex CLI (codex)
  • Aider (aider)

Or configure manually in .kiro/config.json:

{
  "reviewers": ["copilot", "claude", "gemini"]
}

Usage

# Review recent changes (uses configured reviewers)
kspec review

# Review specific target
kspec review "src/auth/*.js"

# Quick review without agentic loop
kspec review --simple

# Build with review loop
kspec build --review

Session Files

Review sessions are logged to .kiro/sessions/ with full transcripts:

.kiro/sessions/review-pr-2026-03-04T20-29-38-940Z.md

Each session captures doer output, reviewer critiques, and final status (APPROVED, NEEDS_CHANGES, or NEEDS_HIL).

Contracts (Beta)

Enforce structured outputs and non-negotiable checks in your spec. This prevents context loss and regression by ensuring specific files and patterns exist before verification proceeds.

Add a ## Contract section to your spec.md:

## Contract

\`\`\`json
{
  "output_files": ["package.json", "src/index.js"],
  "checks": [
    { "type": "contains", "file": "package.json", "text": "\"name\": \"my-app\"" }
  ]
}
\`\`\`

kspec verify will automatically validate these rules.

See Contracts Documentation for full details.

Enterprise Governance

For teams operating under SOC2, regulated environments, or any org with central control over MCP/model usage. All four governance settings are off by default: kspec asks once during kspec init whether to configure them, and the prompt only fires when you opt in.

How to enable

# Option 1 — opt in interactively
kspec init
> Configure enterprise governance? (MCP/model registries, prompt logging, IdP) (y/N): y

# Option 2 — flip the prompt default to Yes for the whole org
export KSPEC_ENTERPRISE=1   # in dev container / shell init
kspec init                  # prompt now defaults Y, devs can still opt out per-project

# Option 3 — non-interactive (CI / org templates)
kspec init --enterprise

What it configures

SettingWhat it doesWhy it matters
MCP registry URLAdmin-hosted JSON allow-list of approved MCP servers. Kiro fetches it every 24h and auto-revokes unapproved serversCentral fleet control, automatic revocation
Model registry URLAdmin-hosted approved-models list. Off-policy model: fields get rewritten to the org defaultCost control + regulatory model whitelisting
Identity providerOkta / Microsoft Entra ID / AWS IAM Identity Center / OtherRequired for the above governance features
Prompt loggingDocuments (in steering) that prompts are recorded by Kiro for SOC2 / regulatory auditSurfaces "don't paste secrets" guidance to every agent

What it ships

Opting in writes .kiro/steering/enterprise-governance.md with inclusion: always, so every agent loads it as context. It substitutes your registry URLs and IdP into the doc and reminds agents to:

  • Cite governance gaps in spec output instead of silently ignoring them
  • Surface model/MCP needs as questions rather than guessing
  • Treat audit.log and .kiro/sessions/ as evidence (don't delete)

See Kiro enterprise governance docs for the upstream feature spec.

CI/CD Integration

kspec init --ci scaffolds two things so kspec runs on every PR:

1. GitHub Actions workflow

Drops .github/workflows/kspec-review.yml:

name: kspec review
on:
  pull_request:
    types: [opened, synchronize, reopened]
jobs:
  kspec-review:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - run: curl -fsSL https://kiro.dev/install.sh | sh
      - run: npm install -g kspec
      - env: { KIRO_API_KEY: ${{ secrets.KIRO_API_KEY }} }
        run: kspec review --simple --trust-tools=read,shell --no-interactive
      # ...posts review as PR comment via actions/github-script

You only need to add KIRO_API_KEY as a repo secret — everything else is wired up.

2. CI hooks preset

.kiro/settings/hooks.json with:

  • preToolUse — audit-log every shell command, hard-block destructive patterns (rm -rf, git push, sudo, curl http)
  • postToolUse — audit-log every file write
  • onSpecComplete — runs kspec verify and kspec sync-jira --progress automatically
  • onSessionStop — refreshes CONTEXT.md so the next CI run starts clean

Powered by Kiro CLI 2.0+ headless mode (--no-interactive, KIRO_API_KEY, --trust-tools). See Kiro headless docs.

Agent Skills (Kiro CLI 2.1+)

In addition to the JSON custom agents (/agent swap kspec-spec), kspec ships Agent Skills as .kiro/skills/<name>/SKILL.md. In Kiro CLI 2.1+ these auto-become slash commands in the default chat agent — no /agent swap required.

> /kspec-spec
> Build a payment processing feature
  (clarifies → spec.md → spec-lite.md)

> /kspec-build
  (executes tasks with strict TDD)

> /kspec-review
  (multi-CLI parallel review)

5 skills shipped: kspec-spec, kspec-build, kspec-review, kspec-verify, kspec-jira. Created when you answer Y to "Create Kiro Agent Skills?" during kspec init (default Yes). See Kiro skills docs.

Agent Permissions (Least-Privilege)

Every kspec agent ships with a toolsSettings block scoping what it can touch:

Agent typeWrite pathsShell scope
State-only (analyse, context, refresh, jira, demo, estimate).kiro/**none
Spec pipeline (spec, design, tasks, revise, spike).kiro/** (+ AGENTS.md for spec)none
Verifiers (verify, review).kiro/**read-only commands + test runners
Code-modifying (build, fix, refactor).kiro/**, src/**, lib/**, test/**, *.ts/.js/.py/.go/.rs, ...npm/pnpm/yarn, pytest, go test, cargo, git status/diff/add/commit

Universal denylist (all agents): .env*, **/secrets/**, **/credentials/**, *.pem, *.key, .git/**, node_modules/**, vendor/**, dist/**, build/**.

Shell denylist (build/verify/review): rm -rf, git push, git reset --hard, sudo, curl, wget, npm publish, pip install, apt.

Subagent delegation graph — every agent declares an explicit availableAgents list (e.g. kspec-build → [verify, review, fix], kspec-verify → [] terminal). Makes the call graph auditable and admin-restrictable. See Kiro custom agents reference.

Powers

Powers are modular knowledge files that enhance AI agent capabilities. kspec ships with 5 powers:

PowerDescription
contractEnforce structured outputs and checks in specs
documentDocumentation best practices (README, CONTRIBUTING, CHANGELOG, ADRs)
tddTest-driven development patterns and workflows
code-reviewCode review checklists and quality standards
code-intelligenceTree-sitter and LSP setup for enhanced AI assistance

In Kiro IDE: Open the Powers panel and install from this repository's powers/ directory.

With kspec CLI: Powers are reference documentation in the powers/ directory. Agents can read them for context.

Custom Powers: Create your own in powers/{name}/POWER.md following the Kiro power format.

Steering Documents

Steering files in .kiro/steering/ are project rules that agents load as context. kspec ships 7 templates with Kiro-native inclusion modes:

FileInclusionWhen it loads
product.mdalwaysEvery prompt
tech.mdalwaysEvery prompt
testing.mdalwaysEvery prompt
security.mdalwaysEvery prompt
api-standards.mdfileMatchWhen **/api/**, **/routes/**, **/handlers/**, **/controllers/** files are touched
frontend.mdfileMatchWhen **/*.tsx, **/*.jsx, **/components/**, **/styles/**, **/*.css files are touched
backend.mdfileMatchWhen **/server/**, **/services/**, **/db/**, **/migrations/** files are touched

fileMatch keeps context small in monorepos — frontend rules don't pollute backend agent prompts.

Non-destructive merge

If a steering file already exists (e.g. from base Kiro IDE init), kspec init merges instead of overwriting:

  • Missing H2 sections are appended with a <!-- added by kspec --> marker
  • Missing frontmatter keys are added; existing values never overwritten
  • Files outside the 7 known templates are left fully untouched

Legacy inclusion_mode: on_demand files are auto-migrated to inclusion: auto on next init.

Context Management

kspec maintains context that survives AI context compression:

.kiro/CONTEXT.md (auto-generated)
├── Current Spec: 2026-01-24-user-auth
├── Progress: 3/12 tasks completed
├── Design: present / not yet created
├── Requirements Summary (from spec-lite)
├── Decisions & Learnings
├── Jira Links (if integrated)
└── Quick Commands (design, tasks, build, verify...)

Agents read CONTEXT.md first, automatically restoring state after context compression. CONTEXT.md is refreshed both before and after agent chat sessions.

kspec context    # View and refresh context (CLI)

Or refresh inline without leaving kiro-cli:

> /agent swap kspec-context

Note: There is no automatic hook for /compact — CONTEXT.md won't auto-refresh on context compaction. Use kspec-context agent or run kspec context after compacting.

Agents & Shortcuts

AgentShortcutPurpose
kspec-analyseCtrl+Shift+AAnalyse codebase, update steering
kspec-specCtrl+Shift+SCreate specifications
kspec-designCtrl+Shift+DCreate technical design from spec
kspec-tasksCtrl+Shift+TGenerate tasks (uses design if present)
kspec-buildCtrl+Shift+BExecute tasks with TDD
kspec-verifyCtrl+Shift+VVerify spec/design/tasks/implementation
kspec-reviewCtrl+Shift+RCode review (+ configured reviewers)
kspec-jiraCtrl+Shift+JJira integration (pull, sync, subtasks)
kspec-fixCtrl+Shift+FFix bugs (abbreviated pipeline)
kspec-refactorCtrl+Shift+GRefactor code (no behavior change)
kspec-spikeCtrl+Shift+IInvestigate/spike (no code)
kspec-reviseCtrl+Shift+ERevise spec from feedback
kspec-demoCtrl+Shift+WGenerate stakeholder walkthrough
kspec-estimateCtrl+Shift+XAssess complexity
kspec-contextCtrl+Shift+CRefresh CONTEXT.md inline
kspec-refreshGenerate AI summary of spec

Switch agents in kiro-cli: /agent swap kspec-build or use keyboard shortcuts.

Every agent includes a PIPELINE section suggesting contextual next steps — so you can navigate the full workflow without leaving kiro-cli.

Context Refresh After /compact

When you run /compact in kiro-cli, CONTEXT.md may become stale. Refresh it inline:

> /agent swap kspec-context

This regenerates CONTEXT.md with current spec progress without leaving your session.

ACP (Agent Client Protocol)

ACP enables Kiro to work with JetBrains IDEs (IntelliJ, WebStorm, PyCharm) and Zed editor.

JetBrains Setup

Create or edit ~/.jetbrains/acp.json:

{
  "agent_servers": {
    "Kiro Agent": {
      "command": "/full/path/to/kiro-cli",
      "args": ["acp"]
    }
  }
}

Replace /full/path/to/kiro-cli with the actual path (find with which kiro-cli). Restart your IDE after configuration.

Once ACP is configured, kspec agents work the same way as in Kiro IDE. Your .kiro/agents/ configurations are loaded automatically.

See: https://kiro.dev/docs/cli/acp/

Code Intelligence

Kiro includes built-in code intelligence powered by tree-sitter, with optional LSP integration for deeper analysis.

/code init      # Index your project
/code status    # Check indexing status

This enables structural code understanding (symbols, references, definitions) for 18 languages. See the code-intelligence power for detailed setup and usage guide.

See: https://kiro.dev/docs/cli/code-intelligence/

Structure

.kiro/
├── config.json           # User preferences (commit)
├── .current              # Current active spec path (local only)
├── CONTEXT.md            # Auto-generated context (local only)
├── memory.md             # Project learnings (commit)
├── specs/
│   └── 2026-01-22-feature/
│       ├── spec.md       # Full specification (commit)
│       ├── spec-lite.md  # Concise (for context compression)
│       ├── design.md     # Technical design (commit, optional)
│       ├── tasks.md      # Implementation tasks (commit)
│       ├── memory.md     # Feature learnings (commit)
│       ├── metadata.json # Spec type metadata (commit)
│       ├── metrics.json  # Pipeline timeline (commit)
│       ├── estimate.md   # Complexity estimate (commit, optional)
│       ├── demo.md       # Stakeholder walkthrough (commit, optional)
│       └── jira-links.json # Jira issue links (commit)
├── milestones/           # Milestone groupings (commit)
├── sessions/             # Review session logs (local only)
├── steering/             # Project rules — incl. enterprise-governance.md if --enterprise (commit)
├── agents/               # kspec-generated agents — *.json (CLI) + *.md (IDE chat, optional) (commit)
├── skills/               # Kiro Agent Skills — /<name> slash commands in default chat (commit)
├── settings/
│   ├── mcp.json          # MCP config (local only)
│   └── hooks.json        # Hooks (basic / enterprise / documentation / ci preset)
└── mcp.json.template     # MCP config template (commit, no secrets)

When kspec init --ci is used, also creates:

.github/workflows/kspec-review.yml   # Headless review on every PR (commit)

Team Collaboration

kspec is designed for team collaboration. Most files should be committed to share specifications, tasks, and guidelines across your team.

What to Commit

PathCommit?Why
.kiro/config.jsonYesProject preferences (incl. enterprise + ideAgents + skills flags)
.kiro/specs/YesSpecifications, designs, tasks, memory
.kiro/steering/YesShared product, tech, testing guidelines (+ enterprise-governance.md if --enterprise)
.kiro/agents/YesAgent configurations — JSON for CLI, optional .md for IDE chat
.kiro/skills/YesKiro Agent Skills — /<name> slash commands
.kiro/mcp.json.templateYesMCP setup template (no secrets)
.kiro/memory.mdYesProject learnings
.github/workflows/kspec-review.ymlYesIf --ci was used — headless review on PRs
.kiro/.currentNoPersonal working state
.kiro/CONTEXT.mdNoAuto-generated, local state
.kiro/audit.logNoCI hooks audit trail (local only)
.kiro/sessions/NoReview session logs
.kiro/settings/NoLocal MCP config + hooks.json
~/.kiro/mcp.jsonN/APersonal secrets in home directory

Setting Up MCP for Teams

kspec init creates .kiro/mcp.json.template which uses mcp-remote with OAuth (no API tokens needed):

{
  "mcpServers": {
    "atlassian": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.atlassian.com/v1/sse"],
      "timeout": 120000
    }
  }
}

Each team member copies to their settings:

mkdir -p ~/.kiro/settings
cp .kiro/mcp.json.template ~/.kiro/settings/mcp.json

Or add via CLI: kiro-cli mcp add --name atlassian

See SECURITY.md for security best practices.

Jira Integration

Bridge the gap between BAs/PMs and developers by integrating with Jira via Atlassian MCP.

Pull Requirements from Jira

kspec spec --jira PROJ-123,PROJ-456 "User Authentication"

This fetches issue details, extracts acceptance criteria, and creates a unified spec with source attribution.

Push Specs to Jira

kspec sync-jira                      # Smart: updates existing or creates new
kspec sync-jira --create             # Force create new issue
kspec sync-jira --project SECOPS     # Create in specific project
kspec sync-jira --update PROJ-789    # Update specific issue

sync-jira is smart — it checks jira-links.json for an existing linked issue and updates it by default. Use --create to force a new issue.

Pull Latest Updates

kspec jira-pull

Fetches the latest state of linked Jira issues, generates a change report, and presents changes for approval before modifying spec.md.

Create Subtasks

kspec jira-subtasks PROJ-789

Generate Jira subtasks from tasks.md for progress tracking.

Prerequisites

Configure Atlassian MCP using one of these methods:

Option 1: Use kiro-cli (recommended)

kiro-cli mcp add --name atlassian

Option 2: Manual configuration

Add to .kiro/settings/mcp.json (workspace) or ~/.kiro/settings/mcp.json (user):

{
  "mcpServers": {
    "atlassian": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.atlassian.com/v1/sse"],
      "timeout": 120000
    }
  }
}

Or add via CLI: kiro-cli mcp add --name atlassian

See: https://kiro.dev/docs/cli/mcp/

See Team Collaboration for secure team setup with environment variables.

Migrating from v1

kspec v2.0 consolidates everything into .kiro/ (previously split between .kspec/ and .kiro/).

When you run any kspec command, it will automatically detect .kspec/ and offer to migrate:

$ kspec status

  kspec v2.0 Migration

  kspec now stores everything under .kiro/ instead of .kspec/

  Files to migrate:
    - config.json
    - specs/ (3 specs)

  Migrate .kspec/ to .kiro/ now? (Y/n):

Migration moves config.json, .current, CONTEXT.md, memory.md, and specs/ from .kspec/ to .kiro/, then removes the empty .kspec/ directory.

Configuration

Set during kspec init:

  • Date format: YYYY-MM-DD, DD-MM-YYYY, or MM-DD-YYYY
  • Auto-execute: ask (default), auto, or dry-run
  • Model: AI model for agents (claude-sonnet-4.6, claude-opus-4.6, claude-haiku-4.5, or custom)
  • Jira project: Default project key for sync-jira (when Atlassian MCP detected)
  • Reviewers: Multi-CLI reviewers for agentic review loop (Copilot, Claude, Gemini, etc.)
  • IDE chat subagents: Optional .md agent files for Kiro IDE chat (default No)
  • Agent Skills: Slash-command skills (/kspec-spec etc.) for default chat (default Yes)
  • Enterprise governance: MCP/model registry URLs, IdP, prompt logging (opt-in, off by default)
  • Hooks preset: none / basic / enterprise / documentation / ci

Environment variables

VariableEffect
KSPEC_ENTERPRISE=1Flips the "Configure enterprise governance?" prompt default to Yes (orgs set this in dev container / shell init)
KIRO_API_KEYRequired for headless mode — used by the --ci GitHub Actions workflow

File Locking

kspec build uses file locking to prevent concurrent builds on the same spec. If another build is running, you'll see an error with the PID and lock time. Locks auto-expire after 30 minutes if the process crashes.

Auto-Updates

kspec checks for updates automatically (cached for 24 hours). Check manually:

kspec update

Or check when viewing version:

kspec --version

Known Limitations

LimitationWorkaround
No /compact hookCONTEXT.md doesn't auto-refresh on context compaction. Run /agent swap kspec-context or kspec context manually.
spec-lite.md auto-update is truncationtruncateSpecLite() truncates spec.md (not AI summary). Run kspec refresh or /agent swap kspec-refresh for AI-generated summary.

Requirements

  • Node.js >= 18
  • Kiro CLI or Amazon Q CLI
  • Atlassian MCP (optional, for Jira integration)

Documentation

License

MIT

Keywords

kiro

FAQs

Package last updated on 13 May 2026

Did you know?

Socket

Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.

Install

Related posts