PixDig MCP server
Commit Nov 4, 2026Lets an AI client search a directory of over 150 vetted AI tools, 40 LLMs, and 1,000 GitHub repositories to find software, models, and installation prompts.
Also known as HelloPixDig/pixdig-mcp
- Stars
- 1
- Tools
- 9
- Downloads, 30 days
- —
- Latest version
- —
- Last commit
- Nov 4, 2026
- License
- MIT
What is PixDig MCP?
PixDig lets an AI client search a directory of over 150 vetted AI tools, 40 LLMs, and 1,000 GitHub repositories to find software, models, and installation prompts. It exposes 9 tools, such as search_tools, get_tool, list_usecases.
Summary generated from the project's README
About PixDig MCP server
Finding vetted software and models for AI development is what PixDig MCP enables for developers using agents like Claude Code, Cursor, or Gemini CLI. It provides direct access to a directory containing over 150 AI tools, 40 large language models, and a curated catalog of 1,000 GitHub repositories. Users can query the server to find specific software requirements, such as hardware compatibility or license types, without leaving their coding environment.
The server operates locally via stdio and requires Node.js 18 or higher. It is a read-only connection that does not require an account, API keys, or any personal credentials. It fetches data through simple HTTPS GET requests to public JSON files, ensuring a small footprint and high transparency.
Developers can use specific tools like search_tools to find software by use case, recommend_models to identify LLMs that fit specific VRAM limits, or find_alternative to locate free versions of paid services. The server also includes a search_catalog tool for exploring GitHub repositories and a featured tool to see current highlights.
What you can do with it
- 1Find hardware compatible modelsAsk your agent to identify open weights models that can run on a specific amount of VRAM.
- 2Get installation instructionsRetrieve official commands and setup prompts to install a specific tool through your agent.
- 3Search curated GitHub repositoriesBrowse a large collection of specialized repositories filtered by theme or popularity.
Written from the project's README. Every use case is checked against a quote from it.
PixDig MCP tools 9
Detected from the READMEsearch_toolsget_toollist_usecasesrecommend_modelsget_modelfind_alternativefeaturedsearch_catalogget_ranking
Automated checks
Ran Nov 14, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachablehellopixdig/pixdig-mcp is public and reachablegithub.com/hellopixdig/pixdig-mcp ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-11-04github.com/hellopixdig/pixdig-mcp ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model9 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNo tool named like delete, drop or execPass
- Remote endpoint answers the MCP handshakeNot checked
README
View on GitHubPixDig MCP
The AI gems directory, right inside your agent.
PixDig MCP connects your coding agent to the PixDig directory: more than 150 vetted AI tools (real GitHub stars, licence, price, compatibility, the machine you need), 40+ LLMs and a curated catalogue of about 1,000 GitHub repositories, with a ready-to-follow install prompt for each tool. Read-only, no account, no API key.
See it work
What you can ask your agent
- “Which free tools can edit videos locally?”
- “Find a tool for transcription that works with Ollama.”
- “Install ComfyUI for me.” → the agent fetches the page, shows you the official commands and the install prompt, and asks before running anything.
- “Which LLM can I run with 12 GB of VRAM?”
- “Find me a free alternative to ElevenLabs.”
- “Which gems are featured right now?”
Install
Requires Node.js 18+ (and git for the GitHub form).
Claude Code
claude mcp add pixdig -- npx -y github:hellopixdig/pixdig-mcp
Codex (~/.codex/config.toml)
[mcp_servers.pixdig]
command = "npx"
args = ["-y", "github:hellopixdig/pixdig-mcp"]
Cursor (~/.cursor/mcp.json) · Gemini CLI (~/.gemini/settings.json)
{
"mcpServers": {
"pixdig": { "command": "npx", "args": ["-y", "github:hellopixdig/pixdig-mcp"] }
}
}
No git? The same package is also served by the PixDig website (no npm account involved):
claude mcp add pixdig -- npx -y https://hellopixdig.com/pixdig-mcp.tgz
Prefer to check first? The whole server is one readable file, src/index.js (about 270 lines, two dependencies).
How it works
Your agent calls the local MCP server over stdio. The server only reads public JSON files published by the PixDig website, refreshed every night.
The 9 tools
| Tool | What it does |
|---|---|
search_tools |
Search by keywords, use case, compatibility (Claude Code, Ollama, MCP…), free, local, open source. Sort by stars or recency. |
get_tool |
Full page of a tool: what it does, recommended machine, requirements, official install commands and a ready-to-paste install prompt. |
list_usecases |
The directory's use cases (video, image, audio, agents, memory…) with counts. |
recommend_models |
LLMs by use case and type (paid / open weights). With memory_gb, only the open models that fit in that memory at 4-bit. |
get_model |
Details of a model: licence, parameters, context, access, required machine, official links. |
find_alternative |
Free or open-source alternatives to a paid service (ChatGPT Plus, Midjourney, ElevenLabs, Zapier, Notion, Cursor…). No argument lists every service covered. |
featured |
The gems currently featured. |
search_catalog |
Search the curated catalogue of about 1,000 GitHub repositories (skills, MCP servers, agents, LLM, voice, media, tools), ranked by stars and freshness. Filter by theme. |
get_ranking |
Ranking of the tools by GitHub stars: last 24 hours (day), 7 days (week) or all time (all). |
Every tool accepts lang: "fr" | "en" | "es".
Trust and transparency
- Read-only. All seven tools are declared read-only. The server runs no command and writes no file.
- One network call type. Plain HTTPS
GETof public JSON fromhellopixdig.com/data/api(or a local folder, see below). The only thing sent is aUser-Agentheader,pixdig-mcp/<version>. - No telemetry, no analytics, no account, no API key.
- Sponsored content is labelled. Paid placements are marked
PARTNERSHIP / SPONSOREDin answers, and the server tells the agent to say so. PixDig sells visibility, never an opinion. - Your agent asks first. The server instructs the agent to show install commands and get your approval before running them.
- Small surface. One file, two dependencies (
@modelcontextprotocol/sdkandzod). Data is cached in memory for one hour.
Configuration
| Variable | Default | Purpose |
|---|---|---|
PIXDIG_DATA |
https://hellopixdig.com/data/api |
Another source: a URL, or a local folder containing tools.fr.json, tools.en.json, tools.es.json, models.json, meta.json, alternatives.json. |
Development
git clone https://github.com/hellopixdig/pixdig-mcp.git
cd pixdig-mcp
npm install
npm test
npm test starts the server over stdio and calls every tool (it reads the live data unless PIXDIG_DATA is set).
Suggest a tool
Built an AI tool? Submit it here: free listing or paid feature, and we reply to every application. A listing is never automatic and never a guarantee.
More
Contributing · Security policy · Changelog
Released under the MIT licence. Made by PixDig, an AI watch run from Limoges, France.
PixDig MCP questions
What tools does PixDig MCP provide?
It exposes 9 tools, including search_tools, get_tool, list_usecases, recommend_models, get_model.
Does I need an API key to use PixDig MCP?
No, the server is read-only and requires no account or API key.
Can the server help me find free alternatives to paid AI services?
Yes, it includes a tool specifically designed to find free or open-source alternatives to services like ChatGPT Plus, Midjourney, and ElevenLabs.
How does the server handle installation commands?
The server provides official install commands and prompts, but instructs the agent to ask for your approval before running anything.
Can I filter LLMs by my hardware specifications?
Yes, you can find open models that fit your specific VRAM capacity.