JustClaw-It MCP documentation
Read plain Markdown
# JustClaw-It MCP documentation
JustClaw-It is a remote MCP server for coding agents. Your agent searches 4,519,159 licence-cleared components mined from real open-source repositories, then pulls the one it wants: full source, a GitHub permalink, the SPDX licence, the copyright holder and an attribution header to keep with the code.
- **Search is free.** `check_component_security`, `find_similar`, `library_stats`, `list_categories`, `list_licenses`, `search_components` cost 0 credits and are rate-limited. `search_components` even works without a key (5 a minute, 50 a day per network); everything else needs a free key or a signed-in connector.
- **A pull costs 1 credit.** `get_component` returns the source. The Pro tools cost more (3 and 10 credits).
- **Plans:** Free ($0: free search, up to 300 pulls a month (10 a day) plus 30 on signup), Solo $10/month, Indie $19/month, Pro $49/month. Every plan is a hard cap: nothing is ever billed beyond the plan price (plus any sales tax shown at checkout).
- **Licence and attribution travel with every component**, and the public library only serves permissive and weak-copyleft code.
- **Claude.ai and ChatGPT:** add `https://justclaw.it.com/mcp/signin` as a connector and sign in with your JustClaw-It account; no API key to paste ([Claude.ai](/docs/clients/claude-ai), [ChatGPT](/docs/clients/chatgpt)).
- **For agents:** 4 free [MCP prompts](/docs/prompts), a one-command [skill + agent pack](/docs/agents-pack), and the free `check_component_security` tool with OSV severity ([Security advisories](/docs/security-advisories)).
New here? Start with the [Quickstart](/docs/quickstart).
## Quickstart
1. [Get a free API key](/signup) (email only, no card).
2. Add the server to your agent. Claude Code, one command:
```sh
claude mcp add --transport http --scope user justclaw-it https://justclaw.it.com/mcp --header "Authorization: Bearer <YOUR_API_KEY>"
```
Cursor, VS Code, Codex, Windsurf, Claude Desktop, OpenClaw and Hermes take one config entry each; see the client setup below. Restart your agent and ask: *"search JustClaw-It for a stripe checkout handler"*.
No key yet? `search_components` also works without one, with a small per-network limit, so you can try it first. Every other tool needs a free key.
On **Claude.ai or ChatGPT**, there is no config file: add `https://justclaw.it.com/mcp/signin` as a custom connector, sign in with your JustClaw-It account and approve the app on the consent screen. Steps: [Claude.ai](/docs/clients/claude-ai), [ChatGPT](/docs/clients/chatgpt).
## Client setup (manual)
Server URL `https://justclaw.it.com/mcp` (Streamable HTTP), header `Authorization: Bearer <YOUR_API_KEY>`. Replace `<YOUR_API_KEY>` with your key. Each client also has its own page with verification steps: [Claude Code](/docs/clients/claude-code), [Claude.ai](/docs/clients/claude-ai) and [ChatGPT](/docs/clients/chatgpt) (sign in, no key), [Claude Desktop](/docs/clients/claude-desktop), [Cursor](/docs/clients/cursor), [Windsurf](/docs/clients/windsurf), [VS Code](/docs/clients/vscode), [Codex](/docs/clients/codex), [OpenClaw](/docs/clients/openclaw), [Hermes Agent](/docs/clients/hermes), [other clients](/docs/clients/other).
### Claude Code
One command:
```sh
claude mcp add --transport http --scope user justclaw-it https://justclaw.it.com/mcp \
--header "Authorization: Bearer <YOUR_API_KEY>"
```
### Cursor
Add to `~/.cursor/mcp.json`:
```json
{
"mcpServers": {
"justclaw-it": {
"url": "https://justclaw.it.com/mcp",
"headers": { "Authorization": "Bearer <YOUR_API_KEY>" }
}
}
}
```
### Codex CLI
Add to `~/.codex/config.toml`:
```toml
[mcp_servers.justclaw-it]
url = "https://justclaw.it.com/mcp"
http_headers = { "Authorization" = "Bearer <YOUR_API_KEY>" }
```
### OpenClaw
Register the server, then reload a running gateway:
```sh
openclaw mcp set justclaw-it '{"url":"https://justclaw.it.com/mcp","transport":"streamable-http","headers":{"Authorization":"Bearer <YOUR_API_KEY>"}}'
openclaw mcp reload
```
### Hermes Agent
Add to `~/.hermes/config.yaml`, and put `JUSTCLAW_IT_API_KEY=<YOUR_API_KEY>` in `~/.hermes/.env`:
```yaml
mcp_servers:
justclaw-it:
url: "https://justclaw.it.com/mcp"
headers:
Authorization: "Bearer ${JUSTCLAW_IT_API_KEY}"
```
### Windsurf
Add to `~/.codeium/windsurf/mcp_config.json`:
```json
{
"mcpServers": {
"justclaw-it": {
"serverUrl": "https://justclaw.it.com/mcp",
"headers": { "Authorization": "Bearer <YOUR_API_KEY>" }
}
}
}
```
### Claude Desktop
Add to `claude_desktop_config.json` (macOS `~/Library/Application Support/Claude/`, Windows `%APPDATA%\Claude\`, Linux `~/.config/Claude/`). Claude Desktop reaches remote servers through the `mcp-remote` bridge, so it needs Node.js:
```json
{
"mcpServers": {
"justclaw-it": {
"command": "npx",
"args": ["-y", "mcp-remote@latest", "https://justclaw.it.com/mcp", "--header", "Authorization:${JUSTCLAW_IT_AUTH}"],
"env": { "JUSTCLAW_IT_AUTH": "Bearer <YOUR_API_KEY>" }
}
}
}
```
### Any other MCP client
Streamable HTTP with a standard bearer header:
```json
{ "mcpServers": { "justclaw-it": { "url": "https://justclaw.it.com/mcp", "headers": { "Authorization": "Bearer <YOUR_API_KEY>" } } } }
```
### VS Code
Add to `.vscode/mcp.json` (or your user `mcp.json`); VS Code asks for the key once and stores it:
```json
{
"inputs": [
{ "type": "promptString", "id": "justclaw-it-key",
"description": "JustClaw-It API key", "password": true }
],
"servers": {
"justclaw-it": {
"type": "http",
"url": "https://justclaw.it.com/mcp",
"headers": { "Authorization": "Bearer ${input:justclaw-it-key}" }
}
}
}
```
## Tools
The table is generated from the server's own `tools/list`, so it matches what your agent sees. Full parameters, examples and result fields: [Tool reference](/docs/tools).
| tool | parameters | credits | plan |
| --- | --- | --- | --- |
| [`search_components`](/docs/tools#search-components) | `query`, `limit` (optional), `category` (optional), `min_composite` (optional), `language` (optional), `license` (optional), `offset` (optional), `max_per_repo` (optional), `exclude_vulnerable` (optional) | 0 | All plans |
| [`find_similar`](/docs/tools#find-similar) | `id`, `limit` (optional) | 0 | All plans |
| [`list_categories`](/docs/tools#list-categories) | `limit` (optional) | 0 | All plans |
| [`get_component`](/docs/tools#get-component) | `id` | 1 | All plans |
| [`check_component_security`](/docs/tools#check-component-security) | `component_ids` (optional), `names` (optional), `packages` (optional) | 0 | All plans |
| [`library_stats`](/docs/tools#library-stats) | none | 0 | All plans |
| [`list_licenses`](/docs/tools#list-licenses) | none | 0 | All plans |
| [`get_component_bundle`](/docs/tools#get-component-bundle) | `component_id` | 3 | Pro |
| [`compose_starter`](/docs/tools#compose-starter) | `prompt`, `stack` (optional) | 10 | Pro |
| [`licence_report`](/docs/tools#licence-report) | `files`, `format` (optional), `private_scopes` (optional) | 0 | All plans |
Each result carries five sub-scores (completeness, code quality, deployability, reusability, strategic value) and a `composite` score from 0 to 10; filter with `min_composite`. See [Scores & ranking](/docs/scores).
## Credits & plans
Searching is free. One credit is one component pull (`get_component`). Free gets 30 credits on day one, then 10 a day that do not carry over. Paid plans refill weekly, keep what you don't spend, and earn up to 1.5x credits the longer you stay. Pro adds component bundles and starter projects. When credits run out, pulls stop until the next refill, renewal or an upgrade; you are never billed per credit. Live prices and checkout: [/pricing](/pricing). More: [Credits & plans](/docs/credits).
| plan | price | credits | fair-use cap | API keys | extras |
| --- | --- | --- | --- | --- | --- |
| Free | $0 | 10 per day, plus 30 on day one; resets daily, no carry-over | none | 1 | none |
| Solo | $10/month | 300 per week; unused credits carry over | 6,000 distinct components a year | 1 | none |
| Indie | $19/month | 1,000 per week; unused credits carry over | 15,000 distinct components a year | 1 | none |
| Pro | $49/month | 5,000 per week; unused credits carry over | 100,000 distinct components a year | 3 | component bundles, starter projects |
## Search limits
Search and the other discovery tools (`check_component_security`, `find_similar`, `library_stats`, `list_categories`, `list_licenses`, `search_components`) cost no credits and are rate-limited so free search can't be abused. They need an API key, except `search_components`, which also works without one at lower per-network limits (5 a minute, 50 a day, up to 5 results a page). A call over a limit gets a readable error instead of results; nothing is charged.
| plan | searches per minute, per key | searches per day, per key |
| --- | --- | --- |
| Free | 20 | 300 |
| Solo | 30 | 2,000 |
| Indie | 30 | 4,000 |
| Pro | 60 | 10,000 |
Per network (client IP, IPv6 /64): 60 searches a minute across all keys, and 1,000 a day across new free accounts. Days reset at 00:00 UTC.
## Licences & attribution
- Every component is third-party open-source code, served under its **original licence**. JustClaw-It sells access to the search and retrieval service (credits), not the code.
- The public library only includes permissive and weak-copyleft licences. Strong copyleft (GPL, AGPL) and code without a clear licence are excluded.
- Each result carries its SPDX licence, copyright holder and repository, and returned source starts with an attribution header. **Keep that header**, and include the licence text where the licence requires it (a `THIRD_PARTY_LICENSES` or NOTICE file works).
- The full list of source repositories, licences and copyright holders: `list_licenses`, or [/licenses](/licenses).
More: [Licences & attribution](/docs/licences).
## FAQ
### How is this different from Context7?
Context7 tells your agent how a library works; JustClaw-It hands it production code that uses it. They complement each other: docs for the API, JustClaw-It for a working, attributed implementation. A longer comparison is on the [blog](/blog/justclaw-it-vs-context7).
### What does a search cost?
Nothing. A pull is 1 credit; `get_component_bundle` is 3 and `compose_starter` is 10. Connecting, listing tools and handshakes are free.
### What happens when I run out of credits?
Pulls stop until your next refill (tomorrow on Free, next week on paid plans) or until you upgrade. Search keeps working within its limits. More answers: [FAQ](/docs/faq) and [Troubleshooting](/docs/troubleshooting).
## All pages
- Start: [Quickstart](/docs/quickstart)
- Clients: [Claude Code](/docs/clients/claude-code), [Claude.ai](/docs/clients/claude-ai), [ChatGPT](/docs/clients/chatgpt), [Claude Desktop](/docs/clients/claude-desktop), [Cursor](/docs/clients/cursor), [Windsurf](/docs/clients/windsurf), [VS Code](/docs/clients/vscode), [Codex](/docs/clients/codex), [OpenClaw](/docs/clients/openclaw), [Hermes Agent](/docs/clients/hermes), [other clients](/docs/clients/other)
- Reference: [Tools](/docs/tools), [Credits & plans](/docs/credits), [Licences & attribution](/docs/licences), [How licensing works](/docs/how-licensing-works), [Scores & ranking](/docs/scores), [Security advisories](/docs/security-advisories), [MCP App view](/docs/mcp-app)
- Tooling: [CLI](/docs/cli), [Skill + agent pack](/docs/agents-pack), [MCP prompts](/docs/prompts), [Agent skills](/docs/skills)
- Help: [Compare: Context7, Exa, Ref](/docs/compare), [FAQ](/docs/faq), [Troubleshooting](/docs/troubleshooting), [Changelog](/docs/changelog)
- For LLMs: [/llms.txt](/llms.txt) (index) and [/llms-full.txt](/llms-full.txt) (every page in one file). Every page is also available as Markdown: add `.md` to its URL.