Scaffold new projects from templates and audit repository security, release, and deployment hygiene using specialized audit primitives.
Also known as starter-series/create-starter, starter-series
- Stars
- 2
- Tools
- 9
- Downloads, 30 days
- —
- Latest version
- 0.5.3
- Last commit
- Nov 4, 2026
- License
- MIT
What is create-starter MCP?
create-starter is an MCP server. Scaffold new projects from templates and audit repository security, release, and deployment hygiene using specialized audit primitives. It exposes 9 tools, such as list_templates, create_project, audit_release.
Summary generated from the project's README
About create-starter MCP server
Scaffolding new software projects and verifying repository hygiene is what developers achieve using create-starter. This tool provides a unified engine for both command-line use and AI-driven workflows through an MCP server. It allows users to generate new projects from six specific templates, such as mcp-server, npm-package, or browser-extension, using Zod-validated inputs and atomic file operations to prevent partial directory creation.
Beyond initial setup, the server acts as a continuous auditing gate for release, continuous delivery, and security standards. It provides specialized tools like audit_release to detect version drift, audit_cd to check publication status across platforms like npm or PyPI, and audit_security to verify core CI primitives including CodeQL and dependency audits. Users can integrate this into Claude Desktop via .mcpb bundles or use it within Claude Code as a plugin. It is designed to help developers graduate from simple code exports to professional GitHub Actions deployments by lifting CI/CD layers into existing repositories without modifying application code.
What you can do with it
- 1Scaffold a new projectGenerate a new TypeScript MCP server project using a predefined template.
- 2Audit repository securityVerify that a repository meets the required security baseline including gitleaks and dependency audits.
- 3Verify deployment synchronizationCheck if a package is correctly published and in sync across multiple destinations like npm and PyPI.
- 4Generate a launch reportRun a comprehensive suite of audits to produce a single document proving the repository is ready for launch.
- 5Add CI/CD to existing reposInject professional CI/CD and maintenance layers into an existing project without rewriting the application logic.
Written from the project's README. Every use case is checked against a quote from it.
How to install create-starter MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall create-starter MCP in Claude Code
$ claude mcp add create-starter -- npx -y starter-seriesInstall create-starter MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"create-starter": {
"command": "npx",
"args": [
"-y",
"starter-series"
]
}
}
}Install create-starter MCP in VS Code
.vscode/mcp.json
{
"servers": {
"create-starter": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"starter-series"
]
}
}
}Install create-starter MCP in VS Code CLI
$ code --add-mcp '{"name":"create-starter","type":"stdio","command":"npx","args":["-y","starter-series"]}'Install create-starter MCP in Claude Desktop
claude_desktop_config.json
{
"mcpServers": {
"create-starter": {
"command": "npx",
"args": [
"-y",
"starter-series"
]
}
}
}Install create-starter MCP in Codex
$ codex mcp add create-starter -- npx -y starter-seriesInstall create-starter MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.create-starter] command = "npx" args = ["-y", "starter-series"]
Install create-starter MCP in Gemini CLI
$ gemini mcp add -s user create-starter npx -- -y starter-seriesInstall create-starter MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"create-starter": {
"command": "npx",
"args": [
"-y",
"starter-series"
]
}
}
}Started by HostedMCP on 16 Nov 2026
We ran this command for starter-series 0.5.3 in an isolated container with no network and no keys. It answered and listed 9 tools.
create-starter MCP tools 9
Read from the server with tools/listlist_templatesList all available Starter Series project templates
create_projectScaffold a new project from a Starter Series template
audit_releaseAudit a local repo for release-readiness against the Starter Series quality bar. Detects matched starter, CHANGELOG drift vs merged PRs, version-bump status, and publish-workflow presence. Read-only; never mutates the repo.
audit_cdCheck whether the local repo's version has been published to its destination registries (npm, PyPI, Open VSX, VS Marketplace, AMO, GitHub Releases). Makes outbound HTTPS requests to public registry APIs; never mutates. Reports per-destination drift (in-sync / needs-publish / local-stale / not-found).
audit_securityAudit a local repo for baseline security CI hygiene against the Starter Series quality bar: gitleaks, CodeQL, dependency audit, license check, --ignore-scripts, Dependabot, secret-scanning hints, claude-code-security-review Action, and claude-security-guidance.md. Read-only.
audit_instructionsCheck local instruction discovery, source topology, exact duplicates/overlap, owner decisions and previous-run delta. Read-only by default; update_state explicitly saves the baseline, never source instructions. No semantic or safety enforcement.
generate_launch_proof_reportRun audit_release, audit_cd, audit_security, and audit_instructions together, then return a client-ready Markdown Launch Proof Report. Read-only by default; set write=true to write launch-proof-report.md or output_path inside the target repo.
add_componentLift a starter's CI/CD layer into an EXISTING repo without re-scaffolding — the remediation half of the audit loop (audit_security/audit_release diagnose; this installs the missing files from the matching starter). Components: ci (.github/workflows/ci.yml), security (codeql.yml + SECURITY.md), dependabot (dependabot.yml + auto-merge), maintenance (stale + weekly health check), or all. DRY-RUN BY DEFAULT: returns a per-file plan (create / identical / skip-exists / overwrite) and writes nothing until dry_run is false. Existing-but-different files are skipped unless force — so the dry-run plan doubles as a drift report against the starter. Never touches app code or secrets-bearing CD workflows.
seed_security_guidanceGenerate a starter `claude-security-guidance.md` at the repo root, tailored to the detected Starter Series template. The file is consumed in-session by Anthropic's Claude Code Security Guidance Plugin (released 2026-05-26) as a guard while Claude writes code. Use `force: true` to overwrite an existing file.
| list_templates | List all available Starter Series project templates | |
| create_project | Scaffold a new project from a Starter Series template | |
| audit_release | Audit a local repo for release-readiness against the Starter Series quality bar. Detects matched starter, CHANGELOG drift vs merged PRs, version-bump status, and publish-workflow presence. Read-only; never mutates the repo. | |
| audit_cd | Check whether the local repo's version has been published to its destination registries (npm, PyPI, Open VSX, VS Marketplace, AMO, GitHub Releases). Makes outbound HTTPS requests to public registry APIs; never mutates. Reports per-destination drift (in-sync / needs-publish / local-stale / not-found). | |
| audit_security | Audit a local repo for baseline security CI hygiene against the Starter Series quality bar: gitleaks, CodeQL, dependency audit, license check, --ignore-scripts, Dependabot, secret-scanning hints, claude-code-security-review Action, and claude-security-guidance.md. Read-only. | |
| audit_instructions | Check local instruction discovery, source topology, exact duplicates/overlap, owner decisions and previous-run delta. Read-only by default; update_state explicitly saves the baseline, never source instructions. No semantic or safety enforcement. | |
| generate_launch_proof_report | Run audit_release, audit_cd, audit_security, and audit_instructions together, then return a client-ready Markdown Launch Proof Report. Read-only by default; set write=true to write launch-proof-report.md or output_path inside the target repo. | |
| add_component | Lift a starter's CI/CD layer into an EXISTING repo without re-scaffolding — the remediation half of the audit loop (audit_security/audit_release diagnose; this installs the missing files from the matching starter). Components: ci (.github/workflows/ci.yml), security (codeql.yml + SECURITY.md), dependabot (dependabot.yml + auto-merge), maintenance (stale + weekly health check), or all. DRY-RUN BY DEFAULT: returns a per-file plan (create / identical / skip-exists / overwrite) and writes nothing until dry_run is false. Existing-but-different files are skipped unless force — so the dry-run plan doubles as a drift report against the starter. Never touches app code or secrets-bearing CD workflows. | |
| seed_security_guidance | Generate a starter `claude-security-guidance.md` at the repo root, tailored to the detected Starter Series template. The file is consumed in-session by Anthropic's Claude Code Security Guidance Plugin (released 2026-05-26) as a guard while Claude writes code. Use `force: true` to overwrite an existing file. |
Configuration
CREATE_STARTER_DEBUG
| Name | Description | |
|---|---|---|
| CREATE_STARTER_DEBUG | — |
Automated checks
Ran Nov 12, 2026Full report →Passed our automated checks. That is not a security audit: read the code and give the server the narrowest access it needs. How checks work
- Source repository linked and reachablePass
- Open source license (OSI)Pass
- Commit or release in the last yearPass
- Package name not close to a popular packagePass
- Remote endpoint uses HTTPSNot checked
- Destructive tools are labelledPass
- No hidden instructions to the modelPass
- No critical advisory in the package or its dependenciesPass
README
View on GitHubcreate-starter
Scaffold and audit Starter Series projects — MCP server, Claude Code skill, and CLI in one package.
Part of: Human-Controlled AI Systems — scaffolding is the easy half. What keeps a shipped repo trustworthy is the audit primitives (audit, audit-cd, audit-security) verifying release, CD, and CI security hygiene against a known bar — gating each merge instead of asking a human to re-check by hand.
Korean documentation is maintained under docs/ko/README.md; the root README stays English-only.
Security checks inspect local workflow definitions and referenced starter-series/.github policy, including composite actions and invoked scripts. Evidence includes source paths and content hashes. Unreadable or unsupported reusable workflows are reported as unverified; this checks configuration, not the latest Actions run or a live vulnerability scan.
Currently implemented
- CLI — package and binary identity are
starter-series;npx starter-series my-package --template npm-packagescaffolds one of 6 public templates with Zod-validated input, atomic rename on success, retry + timeout + 50 MB download cap. - MCP server — nine stdio tools:
list_templates,create_project,audit_release,audit_cd,audit_security,audit_instructions,generate_launch_proof_report,seed_security_guidance,add_component. One binary chooses the mode by argv (positional -> CLI, none -> MCP stdio). - Claude Desktop extension —
.mcpbbundle on every release; drag onto the Claude Desktop settings window. - Claude Code plugin + skill —
/plugin install create-starter@starter-seriesships the MCP server and the conversationalcreateskill together. - MCP Registry metadata —
io.github.starter-series/create-starter; registry submission is gated on the unscoped npm package being live and tarball-verified. audit_release— detects matched starter, version vs last-tag drift, CHANGELOG drift vs merged PRs (git log <tag>..HEAD), publish-workflow kind (release-please / publish-on-tag / auto-release).audit_cd— probes npm, PyPI, Open VSX, VS Marketplace, AMO, GitHub Releases for per-destination publish drift (in-sync / needs-publish / local-stale / not-found / unsupported).audit_security— checks 9 items: 8 core CI primitives (gitleaks with pin check, CodeQL, dependency audit, license check,--ignore-scripts, Dependabot grouped, secret-scanning hint, claude-code-security-review Action) plus the optional repo-authorclaude-security-guidance.md. The 8 core checks gate the HARDENED verdict; this repo passes 8/8 core.audit_instructions/audit-instructions— reviews agent instruction files (AGENTS.md,CLAUDE.md,GEMINI.md, Copilot instructions) for exact same-file duplicates, cross-file surface overlap, and keyword-based risk reminders. Duplicate/overlap findings need review; keyword risk summaries are advisory only and not exhaustive safety or semantic drift detection.proof-report/generate_launch_proof_report— runs release, CD, security, and instruction-review audits together and writes a client-readyLaunch Proof Report. This is the monetizable handoff surface: evidence first, no "certified" claim, exit code 1 unless the repo is actually launch-ready.add_component— the remediation half of the audit loop: lifts a starter's CI/CD layer (ci / security / dependabot / maintenance / all) into an existing repo without re-scaffolding. Dry-run by default with a per-file plan (create / identical / skip-exists / overwrite); refuses a dirty git tree unless forced; never touches app code or secrets-bearing CD workflows. The dry-run plan doubles as a drift report against the starter.- Graduation guide —
docs/graduation-from-vibe-coding.md(+ Korean): five-step path from Lovable/Bolt/v0 exports to GitHub Actions + a real deploy target, using the release/CD/security audit primitives.
Planned
audit_cdsupport for Chrome Web Store, EAS, Railway, Fly, and GHCR. Currently reported asunsupportedbecause those destinations require auth or have no public read API.
Design intent
- One binary, two surfaces. CLI and MCP stdio share one scaffolding engine. Argv decides which surface answers. No duplicated logic for "the same thing called from a human vs an agent".
- Atomic on failure. Extraction happens in a sibling
.<name>-incomplete-<rand>directory and only renames into the final path on success. Network failure, corrupt archive, partial write — none of them leaves a half-scaffolded directory behind. - Audit is first-class. Templates ship a security baseline (gitleaks pinned to SHA, CodeQL, Dependabot grouped,
--ignore-scripts, claude-code-security-review). The audit commands check whether a downstream repo still matches that bar — turning the baseline from a one-time scaffold into an ongoing gate. - Eat your own dogfood. This repo passes
audit_security8/8 core checks (HARDENED); the 9th is the optionalclaude-security-guidance.md. If the tool that audits other repos can't pass its own bar, the bar isn't real. - Read-only outside its sandbox. Downloads are capped (50 MB, 30 s timeout, 3 retries). Relative output paths cannot escape cwd; absolute paths are accepted only as explicit user intent.
git initfailure is logged but non-fatal.
Non-goals
- Full vendor parity in
audit_cd. Destinations without a public read API stayunsupportedrather than reporting confidently-wrong state. - Rewriting app code. The graduation flow lifts CI/CD from the matching starter; it never touches application code.
- A general-purpose project generator. Templates are the 6 public Starter Series templates. New stacks land as new starters, not as flags on
create_project. - Semantic instruction drift or AI safety enforcement.
audit_instructionsis a review aid for exact duplicate/surface overlap and keyword reminders. It is not a semantic similarity engine, runtime guardrail, red-team harness, or exhaustive safety/security linter.
Quick start — CLI
npx starter-series my-package --template npm-package
# Or build from a create-starter checkout:
npm ci
npm run build
node dist/index.js my-package --template npm-package
starter-series — scaffold a project from the Starter Series.
Usage
starter-series <name> --template <id> [options]
starter-series audit [path]
starter-series audit-cd [path]
starter-series audit-security [path]
starter-series check [path] [--instructions]
starter-series audit-instructions [path]
starter-series proof-report [path] [--output <file>] [--stdout]
starter-series seed-security-guidance [path] [--force]
starter-series add-component [path] [--component <g>] [--starter <id>] [--apply] [--force]
starter-series --list
starter-series --help
Options
-t, --template <id> Template ID (see --list)
-d, --description <text> One-line project description
-o, --output-dir <path> Output directory (default: ./<name>)
--no-git Skip "git init" after scaffold
--output <file> proof-report output (default: <path>/launch-proof-report.md)
--stdout print proof-report Markdown; with --output, also writes the file
--component <group> add-component group: ci, security, dependabot, maintenance, all
--starter <id> add-component source starter override
--apply Write the add-component plan (default is dry-run)
--force Overwrite differing component files or guidance
--list List templates and exit
-h, --help Show help and exit
-v, --version Print version and exit
Environment
CREATE_STARTER_DEBUG=1 Emit verbose stderr logs
Available templates
| ID | Stack |
|---|---|
mcp-server |
TypeScript + @modelcontextprotocol/sdk + Zod |
mcp-server-python |
Python + FastMCP |
npm-package |
Jest + ESLint + OIDC publish |
browser-extension |
Chrome/Firefox MV3 |
vscode-extension |
VS Marketplace + Open VSX |
docker-deploy |
any language + GHCR + SSH |
Only the six public templates listed here are available for generation and component downloads. Existing projects based on former templates can still be audited.
Run starter-series --list (CLI) or call list_templates (MCP) for the authoritative, up-to-date list.
Graduating from Lovable / Bolt / v0
Already have a working app on a vibe-coding platform and want to graduate to GitHub Actions + your own deploy target? Read docs/graduation-from-vibe-coding.md (한국어) — a 5-step path that uses audit, audit-cd, and audit-security to diagnose your repo, then lifts CI/CD from the matching starter without rewriting your app code.
Install from source
git clone https://github.com/starter-series/create-starter
cd create-starter
npm install
npm run build
Requires Node.js ≥22.
One-click install in Claude Desktop
Grab the latest .mcpb bundle from the Releases page and drag it onto the Claude Desktop settings window. Claude Desktop unpacks the bundled dist/ and node_modules/ and registers create-starter as an MCP server — no npm, no config file, no absolute path.
.mcpb(MCP Bundle, formerly.dxt) is Anthropic's packaged extension format for MCP servers. See Desktop Extensions.
To rebuild the bundle locally:
npm ci
npm run bundle:mcpb # produces create-starter-<version>.mcpb
Use as MCP server
Register the built binary in your MCP client (Claude Desktop, Cursor, etc.):
{
"mcpServers": {
"create-starter": {
"command": "node",
"args": ["/absolute/path/to/create-starter/dist/index.js"]
}
}
}
Then ask your agent: "Use create-starter to scaffold a new discord bot named my-bot." The agent will call list_templates if needed and then create_project.
The binary speaks MCP stdio when called with no extra arguments, and switches to CLI mode when given any positional argument or flag. Both modes share the same scaffolding engine.
Use as Claude Code plugin
The plugin bundles both the MCP server and the create skill — one install wires them up together.
From the Claude Code REPL:
/plugin marketplace add starter-series/create-starter
/plugin install create-starter@starter-series
Then ask Claude: "scaffold a new discord bot named my-bot" and the create-starter:create skill guides the conversation into the MCP tools.
For local development (no marketplace round-trip):
claude --plugin-dir /path/to/create-starter
Point at a git clone so edits in skills/create/SKILL.md or dist/index.js take effect the moment the session starts.
Use via MCP Registry
After registry publication, this server uses the Official MCP Registry namespace:
io.github.starter-series/create-starter
MCP-compatible clients that integrate registry discovery can install it by name without manual path wiring. The registry entry points at the npm package starter-series, so the registry step must run only after the npm package is published and verified.
Ownership is verified through GitHub OIDC (namespace io.github.starter-series/*) and npm tarball inspection (package.json#mcpName). See .github/workflows/publish-mcp-registry.yml for the publish flow.
For npm release setup (trusted-publisher registration, including the post-2026-05-20 allowed-action step), see docs/RELEASING.md.
Tools
Scaffolding:
list_templates— returns the full template table as JSON.create_project— args:template(required) — template ID from the table above.name(required) — project name matching^[A-Za-z0-9][A-Za-z0-9_-]*$.description(optional) — one-line description.output_dir(optional) — defaults to./<name>relative to the MCP server's cwd. Relative paths must stay inside cwd; absolute paths are accepted as explicit user intent.init_git(optional, defaulttrue) — rungit initafter scaffold.
Audit (each takes an optional path arg, default = MCP server cwd; all read-only):
audit_release— release-readiness diagnosis. CLI mirror:starter-series audit [path].audit_cd— per-destination publish-drift probe. CLI mirror:starter-series audit-cd [path].audit_security— baseline CI security hygiene check. CLI mirror:starter-series audit-security [path].audit_instructions— agent-instruction duplicate and surface-overlap review, with advisory keyword risk summaries. CLI mirror:starter-series audit-instructions [path].generate_launch_proof_report— combined Markdown launch handoff from release, CD, security, and instruction-review audits. CLI mirror:starter-series proof-report [path] [--output <file>] [--stdout].seed_security_guidance— generate a starter-awareclaude-security-guidance.mddraft. CLI mirror:starter-series seed-security-guidance [path] [--force].add_component— propose or apply starter CI/CD components to an existing repo as a dry-run plan. CLI mirror:starter-series add-component [path] [--component <g>] [--starter <id>] [--apply] [--force].
Safety & reliability
- Project names are regex-validated before any filesystem touch; relative output paths are rejected if they escape the working directory.
- Downloads enforce a 30 s timeout, 3-attempt exponential backoff, and a 50 MB size cap.
- Extraction happens in a sibling
.<name>-incomplete-<rand>dir; on any failure (network, corrupt archive, extraction error) the tmp dir is removed. The final path only appears via an atomicrenameonce everything succeeded. git initfailures are logged to stderr but do not fail the scaffold; the project is usable without a.gitdirectory.
Supply-chain security pre-wired
Every Starter Series template ships with the 9 checks audit_security looks for — no opt-in required:
| Check | What it catches |
|---|---|
| gitleaks (SHA256-pinned manual install) | Committed secrets in code or history |
| CodeQL (weekly + PR) | Static analysis for JS/TS/Python |
Dependency audit (npm audit --audit-level=moderate / pip-audit) |
Known CVEs in transitive deps |
| License check | GPL/AGPL contamination |
--ignore-scripts on every npm/pnpm/yarn install |
Malicious postinstall scripts |
| Dependabot grouped updates | Lockfile-conflict storms from one-by-one bumps |
| GitHub secret scanning + push protection | Tokens leaked at push time |
anthropics/claude-code-security-review Action on PR |
AI-based diff review |
claude-security-guidance.md (this is the only one you write) |
Org-specific rules consumed by Anthropic's in-session Claude Code Security Guidance Plugin (released 2026-05-26) |
This was Vercel's stack during their 2026-04-21 npm supply-chain incident — they pre-empted compromise via the same pre-wired checks plus Socket/npm/GitHub coordination. The Starter Series ships those checks pre-wired in every starter.
License
MIT © heznpc
Instruction checks
starter-series check --instructions [path]
starter-series check --instructions [path] --json
starter-series check --instructions [path] --update-state
check without --instructions runs the existing release, CD, security and instruction audits. audit-instructions remains a compatible alias. MCP clients keep using audit_instructions(path?, update_state?); the library entry remains auditInstructions(path?, { updateState? }).
One scan covers AGENTS/CLAUDE/GEMINI, overrides, Copilot instructions, Cursor rules, Claude rules/skills and .agents rules/skills/workflows. The check reports import and symlink aliases, verbatim mirrors, local overrides, exact duplicates and cross-file overlap. Canonical selection is inferred from links and filenames unless explicitly configured. It does not detect semantic similarity. Instruction source files are never rewritten.
Optional .starter-series/instructions.json selects a canonical source and records owner decisions. Copy the exact id, evidenceHash and paths from the JSON report into a decision's id, evidenceHash and scope; then explicitly supply the owner's reason, identity and approval date:
{
"schemaVersion": 1,
"canonical": "AGENTS.md",
"decisions": []
}
Each decision requires reason, approvedBy, approvedAt (YYYY-MM-DD), and at least one of reviewAfter, expiresAt (YYYY-MM-DD) or reviewReason. A decision only accepts its exact finding and scope. Evidence changes, expiry or a due review make it stale; approval is never generated by the checker. Accepted findings remain visible. This file records owner approval; it is not an authentication mechanism.
Checks are read-only by default. --update-state (or MCP update_state: true) explicitly saves .starter-series/instructions-state.json for new/changed/known/resolved delta comparison. State contains identities and evidence hashes, not instruction text; commit it only if the team wants a shared baseline. Malformed metadata fails the check. Exit codes remain 0 for clean/advisory or accepted findings, 1 for pending/stale review findings, and 2 for usage or operational errors.
create-starter MCP questions
How do I add create-starter MCP to Claude Code?
Run claude mcp add create-starter -- npx -y starter-series in your terminal. Add --scope user to make it available in every project.
How do I add create-starter MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
What tools does create-starter MCP provide?
It exposes 9 tools, including list_templates, create_project, audit_release, audit_cd, audit_security.
How does the security audit work?
The audit checks nine items, including eight core CI primitives like gitleaks, CodeQL, dependency audits, and license checks, plus an optional security guidance file.
Can I add CI/CD layers to an existing repository?
Yes, the add_component tool allows you to lift CI/CD layers like security, dependabot, or maintenance from a starter into an existing repository without re-scaffolding.