Claude Kit
A Claude Code starter kit built for Azure DevOps teams — agents, hooks, MCP servers, and slash commands that automate the full work item → branch → PR → release → deploy lifecycle. Installed into any project with a single
npxcommand.
Power your agent
पब्लिशर की अपनी कमांड। PlugMCP एंडपॉइंट नहीं।
Host a copy with our keys and logs.
npx -y @chris1807/[email protected]Claude Code में लोकल कमांड के लिए इंस्टॉल लिंक नहीं है। नीचे दी गई कमांड कॉपी करें और टर्मिनल में चलाएँ।
claude mcp add chris1807-claude-kit -- npx -y @chris1807/claude-kit@2.3.10टर्मिनल में चलाएँ।
{ "mcpServers": { "chris1807-claude-kit": { "command": "npx", "args": [ "-y", "@chris1807/[email protected]" ] } }}claude_desktop_config.json
रिमोट URL या आपकी मशीन पर कमांड? दोनों इंस्टॉल में क्या अंतर है
PlugMCP पर चलाएँ
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
लॉन्चपब्लिशर से
A Claude Code starter kit built for Azure DevOps teams — agents, hooks, MCP servers, and slash commands that automate the full work item → branch → PR → release → deploy lifecycle. Installed into any project with a single
npxcommand.
Claude Code starter kit for Azure DevOps teams — agents, hooks, MCP servers, slash commands, and end-to-end work item → PR → release → deploy workflow automation
Claude Kit installs a complete AI development infrastructure into any project that uses Azure DevOps for work tracking, source control (Azure Repos), and CI/CD (Azure Pipelines). It includes specialized agents that handle specific tasks (deployment, database ops, code review), security hooks that prevent mistakes automatically, MCP server connections to your tools (Azure DevOps, databases, Teams, Stripe, Azure CLI), and slash commands that automate the entire workflow from work item to pull request.
The slash commands assume Azure DevOps as the system of record:
AB#1234 ids throughout)Release #N) and tags; CD runs on Azure PipelinesNot using Azure DevOps? The Claude Code primitives (agents, hooks, memory) are still useful, but the slash commands and the deployment workflow won't apply out of the box — you'd need to rewrite the
/implement,/review,/deploy,/create-release,/deploy-release,/cherry-pick,/promote,/rollback,/track,/rework, and/resolve-feedbackcommands against GitHub / GitLab / Jira / etc.
Every session Claude learns from your feedback and gets better at helping you specifically. The infrastructure is modular — install only what your project needs.
npx @chris1807/claude-kit --help
You should see the help output. The package is published publicly on npmjs.com — no authentication or .npmrc setup required.
Every install mode does two things:
~/.claude/agents/ (works across ALL your projects)Best for first-time setup or when you want to pick exactly what you need.
npx @chris1807/claude-kit init
You'll be asked:
Same as Option 1, but you specify the project path upfront:
npx @chris1807/claude-kit init /path/to/your/project
Installs all components. Only asks which database type (there's no sensible default).
npx @chris1807/claude-kit init --all
This installs:
Specify the database type as a flag — no prompts at all:
npx @chris1807/claude-kit init --all --db=mongo
npx @chris1807/claude-kit init --all --db=mssql
npx @chris1807/claude-kit init --all --db=azuresql
npx @chris1807/claude-kit init --all --db=postgres
You can also target a specific directory:
npx @chris1807/claude-kit init /path/to/project --all --db=mongo
Just installs the 13 global agents to ~/.claude/agents/. No project files, no prompts.
npx @chris1807/claude-kit init --global-only
Use this when you just want the global agents on a new machine and will install project files separately per project.
Safe to run as many times as you want. The installer:
--all mode)Already installed on one project? Just run the same command for the next one:
npx @chris1807/claude-kit init /path/to/another/project
The installer automatically skips global agents that are already installed (they're identical) and only installs the project-specific files. You don't need --global-only or any special flag — it just works.
📦 Global Agents → ~/.claude/agents/
= azure-ops.md (identical, skipped) ← already installed, skipped
= security-auditor.md (identical, skipped)
= reviewer.md (identical, skipped)
📦 Project Agents → .claude/agents/ ← fresh install for this project
✓ deployer.md
✓ db-admin.md
...
~/.claude/agents/ ← Global (all projects)
├── backend.md # .NET/C# backend developer (Clean Architecture)
├── frontend.md # React/TypeScript frontend developer
├── legacy.md # Lucee/CFML legacy app developer (RBWO + others)
├── mockup.md # HTML mockup designer
├── reviewer.md # Code reviewer (read-only)
├── test-runner.md # Test executor — xUnit, Vitest, Playwright (read-only)
├── qa.md # Browser QA + bug repro on a running environment (read-only)
├── build-validator.md # Build checker (read-only)
├── lint-checker.md # ESLint + dotnet format checker
├── azure-ops.md # Azure infrastructure management
└── security-auditor.md # Security scanning (read-only)
your-project/ ← Project-specific
├── .claude/
│ ├── agents/
│ │ ├── deployer.md # Commit → push → deploy → monitor
│ │ └── db-admin.md # Database queries and data management
│ │
│ ├── hooks/
│ │ ├── sensitive-data-blocker.sh # BLOCKS mongosh queries for TIN/SSN/PII
│ │ ├── sensitive-data-mcp-blocker.sh # BLOCKS MCP DB queries for TIN/SSN/PII
│ │ ├── sensitive-data-output-blocker.sh # BLOCKS output containing PII fields
│ │ ├── secret-blocker.sh # BLOCKS hardcoded secrets before write
│ │ ├── protected-files.sh # BLOCKS/warns on critical file edits
│ │ ├── auto-format.sh # Auto-runs formatters after edits
│ │ ├── test-on-change.sh # Suggests related tests after edits
│ │ ├── work-item-intent.sh # Routes "create a bug/story" prompts to /create-work-item
│ │ ├── uat-reminder.sh # Reminds to run UAT after features
│ │ └── self-improve.sh # Saves learnings after each session
│ │
│ ├── commands/
│ │ ├── implement.md # /implement AB#1234
│ │ ├── autopilot.md # /autopilot AB#1234
│ │ ├── review.md # /review 142
│ │ ├── deploy.md # /deploy "commit message"
│ │ ├── create-release.md # /create-release 23
│ │ ├── deploy-release.md # /deploy-release 23 staging
│ │ ├── cherry-pick.md # /cherry-pick AB#1234 prod
│ │ ├── promote.md # /promote main dev
│ │ ├── rollback.md # /rollback AB#1234 prod
│ │ ├── add-to-release.md # /add-to-release 24 AB#4599
│ │ ├── track.md # /track release 24
│ │ ├── cleanup-branches.md # /cleanup-branches
│ │ ├── close-orphan-tasks.md # /close-orphan-tasks
│ │ ├── quote.md # /quote AB#1234
│ │ └── explain.md # /explain AB#1234
│ │
│ └── settings.json # Hook and MCP registration
│
├── .mcp.json # MCP server configuration
└── CLAUDE.md # Gets workflow section appended
cd /path/to/your/project
claude
Verify MCP servers are connected:
/mcp
The kit's agents normally run as sequential delegation — the main loop hands work to one specialist at a time through the Task tool. Ultracode is a separate, harness-level gear that authorizes the Workflow tool to fan out many agents in parallel under a deterministic script. The two compose: ultracode doesn't replace the pipeline, it parallelizes the parts that are embarrassingly parallel.
Ultracode is opt-in. It's on only when you type ultracode in a prompt, when a system-reminder confirms it, or when a slash command's instructions say to use Workflow. When it's off, everything runs the normal sequential way — nothing changes.
Only the main Claude Code loop and slash commands can call Workflow. Subagents (anything running under Task) cannot — so ultracode-scale fan-out is a main-loop / slash-command concern.
Kit operations that benefit from ultracode when it's on:
Short, single-query operations (/track, /explain, /close-orphan-tasks) don't need ultracode — they're already one pass and gain nothing from fan-out. Reach for it when the work-list is large and the per-item work is independent.
/implement AB#1234
Claude automatically:
/promote, /cherry-pick, and /deploy-release list it for every environment. Data fixes are planned as run-once migrations the app applies by itself wherever the project has that mechanism; a hand-run script is the fallback and the plan says whyImplementing a whole Feature: point /implement at a Feature work item and it implements the child User Stories in waves driven by the Custom.Order field — stories with the same order value are implemented in parallel (one agent per story, each in an isolated git worktree), and the next order value starts only after the previous wave is merged and green. Everything lands on one feature/ branch with a single review, UAT pass, and PR that links every story.
/autopilot AB#1234
Runs every step of /implement — plan, architect review, branch, Task, implementation, build, tests, lint, code review, PR — but never stops to ask. Where /implement would wait for you, Claude makes the call and logs it. An independent architect-review agent stands in for your plan approval, and the verified find → refute review stands in for your fix decisions. The PR is where you come in: it's labeled autopilot, its first comment thread holds the plan and every decision, and its description carries the UAT checklist nobody has run yet. Autopilot never merges, deploys, or moves an item past Code Review.
It only stops before writing code (item not found, already has a PR, nothing to build, worktree locked by another session). If it can't get the build, tests, or review clean in three rounds, it opens a draft PR with what's left and leaves the item Active. Set /effort before starting, and run in auto mode (or /add-dir the worktree) so Claude Code's own permission prompts don't stall it.
/review 142
Claude automatically:
/deploy "Add payment export feature"
Claude automatically:
dev, test, staging, prod) — never on main, which has no pipeline/create-release 23
Claude automatically:
Release #23 iteration in Azure DevOpsrelease-23/deploy-release 23 staging
Claude automatically:
Ready for Staging here)release/23-to-stagingstaging branch — merging it triggers the Staging pipelineCOM, PAY, …) — candidates are the non-developer names already on those work itemsdev, test, and staging only; a prod PR you merge yourselfStaging and assigns them to the approved verifiersThe same command walks a release through test → staging → prod; with no environment given it picks the next one.
/cherry-pick AB#1234 AB#1235 prod
Cherry-picks specific work items to an environment without a formal release. Same gate check, same verifier question, and the same automatic state advance + assignment once the pipeline is green as /deploy-release. Targeting dev, test, or staging it also merges the PR itself once Serena approves it; a prod PR waits for a person. main is also a valid target — that's how a hot fix that went straight to prod is brought back into the compare branch.
/promote main dev
/promote staging prod
/promote ← auto-detects from the current branch
Creates a PR to promote all code from one branch to the next in the chain (main → dev → test → staging → prod). Shows every commit and work item before confirming, gate-checks the work items, asks who verifies each product group, and advances and assigns them once the merge's pipeline is green. Below production it also merges the PR for you: it polls until Serena (the AI reviewer) approves, then completes it with autocomplete — a PR into prod is always merged by a person. main → dev is the usual first hop after feature PRs merge — nothing has deployed before that.
/rollback AB#1234 prod
/rollback last staging
Reverts specific commits or the last deployment on an environment. Creates a revert branch and PR.
/add-to-release 24 AB#4599 AB#4600
Adds work items to an existing release — assigns them to the iteration and tags them.
/track release 24
/track pipeline
/track AB#4521
/track staging
/track
Shows the status of a release, pipeline, work item, environment, or a high-level overview of everything.
/quote AB#1234
Displays a work item as a formatted blockquote — title, type, state, assignee, description, acceptance criteria, and child items. Useful for pasting into PR descriptions, Teams messages, or discussions.
/explain AB#1234
Reads a work item and explains it in plain language:
/cleanup-branches
/cleanup-branches --dry-run
Finds and deletes branches that have been fully merged. Protects environment branches. Use --dry-run to preview without deleting.
main is the compare branch — the default branch, the one feature PRs target, and the one every environment is compared against. Merging into main deploys nothing. Each environment has its own branch, and a PR merging into that branch is what triggers its CD pipeline:
Note: A project that hasn't migrated yet (e.g. one that still has only
mainanddevelop) keeps working — every command reads the project's Pipeline Configuration table and uses the branch you're on. No branch names are hardcoded.
Every environment has a work item state, and the promotion commands keep them in sync. The "Ready for …" states are human sign-offs; the commands never set them.
/deploy-release, /cherry-pick, and /promote check the gate before creating the PR, ask who verifies each product group while the PR is in review, then watch the PR and the environment's CD pipeline — on a green pipeline they set the state and the assignee together. A failed pipeline changes nothing. /track flags anything whose state lags the branch it's on.
When you run /implement, a branch is automatically created based on the Azure DevOps work item type:
The branch is always created off the current branch — no assumptions are made about which branch you're on.
Note: The Azure DevOps work item type is "Hot Fix" (two words), but the branch prefix and PR label use
hotfix(one word, lowercase).
Code flows through environments via PRs, never by direct push. Every merge after main deploys:
feature/AB#1234-... ──PR──▸ main ──PR──▸ dev ──PR──▸ test ──PR──▸ staging ──PR──▸ prod
(work branch) (compare, (Dev) (Test) (Staging) (Production)
no deploy)
This section walks through the complete workflow from picking up a work item to deploying to production.
Start by switching to your project's development branch and running /implement:
cd /path/to/your/project
claude
# Start from the compare branch
git checkout main
# Implement the work item
/implement AB#1234
Claude will:
../your-project-AB1234). Every later step runs there, so other Claude sessions in the same repo aren't disturbedstory/AB#1234-user-can-view-history) and implement using backend and/or frontend agentsmain, close the child Task with its hours, and move the work item to Code Review/rework recreates it from the branchSay /implement AB#1234 work in place to skip the worktree and work in the current folder, as before.
Add --opus, --sonnet, or --haiku to run every subagent on one model instead of the default Sonnet split: /implement AB#1234 --opus. /rework and /autopilot take the same flag. The main loop stays on your session model, so switch with /model opus first if you want the whole run on Opus.
The PR merges into main. That merge deploys nothing — main is the compare branch. The work reaches Dev when main is promoted:
/promote main dev
Claude asks who verifies each product group on Dev, waits for Serena to approve the PR and merges it, watches the Dev pipeline, and once it's green moves the work items to Ready for Testing and assigns them to those people. (Dev, Test, and Staging merge themselves this way; a production PR you merge yourself.)
For smaller changes that don't need the full /implement workflow:
/deploy "Fix typo in dashboard header"
This commits, pushes, and triggers the pipeline if you're on an environment branch (dev, test, staging, prod). On main or a feature branch it just pushes — nothing deploys until the work is promoted.
Once multiple work items are merged to main and smoke-tested on Dev, group them into a release:
/create-release 23
Claude will:
AB#1234, AB#1235, AB#1236all Ready for Testing user storiesall items tagged sprint-5Release #23 iteration in Azure DevOpsrelease-23/deploy-release 23 test, then staging, then prod/deploy-release 23 test
Claude will:
release-23 and check each is at the gate for Test (Ready for Testing)devrelease/23-to-testrelease/23-to-test → testTesting and assign them to the approved verifiersAfter the PR is reviewed and merged, the Test CD pipeline triggers automatically. QA tests there and sets each passing item to Ready for Staging; then /deploy-release 23 staging repeats the process (gate Ready for Staging, state after merge Staging).
QA and stakeholders test on the Staging environment. If issues are found, fix them with /implement and add the fixes to the release.
When stakeholders have verified on Staging and set the items to Ready to Deploy:
/deploy-release 23 prod
Same process — gate Ready to Deploy, cherry-picks from staging into a PR targeting prod. You merge this one — Claude auto-merges only into dev, test, and staging. After merge the Production CD pipeline triggers; when it comes back green the items move to Deployed and are assigned to whoever you named for each product group.
If staging has 5 user stories but only 3 are Ready to Deploy:
Option A: Create a smaller release
/create-release 24
Include only the 3 ready stories, then /deploy-release 24 prod.
Option B: Cherry-pick specific items
/cherry-pick AB#1234 AB#1235 AB#1236 prod
This cherry-picks just those 3 work items without creating a formal release.
To promote all code from one branch to the next (no cherry-picking):
/promote main dev
/promote staging prod
This creates a PR from the source branch to the next one in the chain containing everything. main → dev is the everyday first hop; staging → prod is for when everything on staging is approved.
You can also auto-detect the next environment:
/promote
If you're on main, it auto-detects main → dev; on staging, staging → prod.
For critical production issues:
git checkout prod
/implement with the Hot Fix work item:
/implement AB#9999
hotfix/AB#9999-fix-crash-on-submit branchprod directly with a hotfix label — merging it deploys straight to productionmain (a second PR from the same branch, or /cherry-pick AB#9999 main) so the next main → dev promotion doesn't overwrite itIf a deployment causes issues:
Revert specific work items:
/rollback AB#1234 prod
Revert the most recent deployment:
/rollback last staging
Claude will:
revert/2026-09-06-on-prod)git revert on each commitMerge the PR to deploy the rollback. Work item states are left alone — Claude tells you which items were reverted so you can decide. If the reverted commits are also on main, the next promotion brings them back unless the revert is applied there too.
For any open PR:
/review 142
Claude reviews for:
any types in TypeScript (should be properly typed)These run automatically — no action needed:
Each team member sets their own. Never commit these.
Add to ~/.zshrc (Mac) or System Environment Variables (Windows):
# MongoDB (if your project uses MongoDB)
export MONGODB_CONNECTION_STRING="mongodb+srv://user:[email protected]/"
# SQL Server (if your project uses SQL Server or Azure SQL)
export MSSQL_CONNECTION_STRING="Server=localhost;Database=MyDb;User Id=sa;Password=...;"
# PostgreSQL (if your project uses PostgreSQL)
export POSTGRES_CONNECTION_STRING="postgresql://user:password@localhost:5432/mydb"
# Stripe (if your project uses Stripe payments)
export STRIPE_SECRET_KEY="sk_test_..."
# Azure CLI (no env var needed — just log in)
az login
# QA sign-in for /qa and /repro — OPTIONAL, per developer. Leave unset and you sign
# in yourself in the browser window (recommended for your own admin account, since a
# password the agent types lands in the local session transcript). The NAMES come
# from your project's `## Environment URLs` section; _TEST / _DEV / _LOCAL / _STAGING
# versions override the plain one on that environment.
export COMPASS_QA_ADMIN_USERNAME="[email protected]"
export COMPASS_QA_ADMIN_PASSWORD="..."
export COMPASS_QA_ADMIN_PASSWORD_TEST="..." # only if your test password differs
# Microsoft Teams (no env vars needed — uses Microsoft Graph device-code auth;
# first invocation prints a code + URL to sign in)
ChannelMessage.SendChannelMessage.Read.AllChat.ReadWriteTeam.ReadBasic.AllTEAMS_* environment variables aboveClaude maintains persistent memory across sessions in ~/.claude/projects/.../memory/. The self-improvement hook prompts Claude to save learnings after each session.
After installing, start a Claude Code session and say:
Set up my memory profile. I'm [your name], a [your role] at [your company].
I work on [your projects]. I prefer [your preferences].
Claude creates the initial memory files. Each subsequent session adds to them automatically.
Create .claude/agents/your-agent.md in your project:
---
name: your-agent
description: What this agent does
tools:
- Bash
- Read
- Write
---
Instructions for the agent...
.claude/hooks/your-hook.sh and make it executable.claude/settings.json under PreToolUse, PostToolUse, or StopCreate .claude/commands/your-command.md:
Do something with $ARGUMENTS.
1. Step one
2. Step two
Use it: /your-command some-argument
Delete any agent, hook, command, or MCP server you don't need. Everything works independently.
When the infrastructure package is updated:
# npx always fetches the latest version
npx @chris1807/claude-kit init /path/to/project
The installer detects existing files and asks whether to overwrite or skip each one.
cd claude-kit
# Make your changes, then bump the version
npm version patch # 2.0.0 → 2.0.1
# Push the tag — GitHub Actions publishes to npm automatically
git push && git push --tags
The
.github/workflows/publish.ymlworkflow runs on any pushedv*tag and publishes to npm using theNPM_TOKENrepo secret. You should not runnpm publishfrom a dev machine.
.mcp.json only contains ${ENV_VAR} references.claude/settings.local.json is gitignored — personal permissions stay privatenpm ERR! 404 Not FoundIf npx @chris1807/claude-kit reports 404, your local npm registry may be set to a private feed that overrides the public one. Check with npm config get registry — it should be https://registry.npmjs.org/.
Verify the settings file is loaded:
claude
/config
Check that .claude/settings.json shows your hooks.
Check that the environment variable is set:
echo $MONGODB_CONNECTION_STRING # Should show your connection string
echo $STRIPE_SECRET_KEY # Should show sk_test_...
az account show # Should show your Azure subscription
MIT
Claude Code starter kit for Azure DevOps teams — agents, hooks, MCP servers, slash commands, and end-to-end work item → PR → release → deploy workflow automation
{ "mcpServers": { "chris1807-claude-kit": { "command": "npx", "args": [ "-y", "@chris1807/[email protected]" ] } }}टूल
अभी टूल सूची नहीं है। रिमोट सर्वर को पहले हेल्थ चेक के बाद सूची मिलती है।
FAQ
इस पेज पर Cursor, Claude Desktop, Claude Code या VS Code के लिए stdio कॉन्फ़िग कॉपी करें। यह पब्लिशर का इंस्टॉल इस्तेमाल करता है, PlugMCP प्रॉक्सी नहीं।
इस रिकॉर्ड में कोई ज़रूरी एनवायरनमेंट वेरिएबल दर्ज नहीं है।
@chris1807/claude Kit stdio पर लोकल चलता है।
पिछली जाँच 2026-11-10 को हुई। Claude Code starter kit for Azure DevOps teams — agents, hooks, MCP servers, slash commands, and end-to-end work item → PR → release → deploy workflow automation