By adorosario
Give Claude Code persistent semantic search over your entire project — even 10,000 files — using CustomGPT.ai's RAG engine. Index your codebase, documents, or any file collection once, then ask questions across everything instantly. No context overflow, no re-reading files. Use /create-agent to index a folder, /query-agent to search it, and /refresh-agent to keep it current. Requires only curl — no Node.js, no Python, no dependencies.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Look up whether a specific file or URL was successfully crawled and indexed in the CustomGPT.ai agent. Returns status, page ID, and diagnostic guidance.
Show the current status of the CustomGPT.ai agent — project metadata, chat availability, and document crawl/index statistics. Use to verify the index is ready before querying.
Create a new CustomGPT.ai agent for a local folder and upload all eligible files into its knowledge base. Saves agent binding to .customgpt-meta.json.
Permanently delete a CustomGPT.ai agent and all its indexed content, then remove the local .customgpt-meta.json binding. Irreversible — requires confirmation.
Permanently remove a specific document (page) from the CustomGPT.ai agent's knowledge base by page ID. Requires confirmation. Use /list-pages to find page IDs.
A Claude Code plugin that gives Claude Code persistent semantic search over any project using CustomGPT.ai's RAG engine. Index your codebase, documents, or any file collection once, then ask questions across everything instantly — no context overflow, no re-reading files.
No build step. No Node.js. No Python. Requires only curl.
Each project maps to one CustomGPT.ai agent. The agent ID and indexed folder are stored in .customgpt-meta.json at your project root. All skills read this file automatically — you never need to pass an agent ID manually.
Files are uploaded to CustomGPT.ai's servers, where they are chunked, embedded, and indexed. Queries run against that index and return AI-generated answers with source citations.
| Requirement | How to get it |
|---|---|
| CustomGPT.ai account | Sign up at app.customgpt.ai |
| API key | app.customgpt.ai/profile#api-keys |
| Claude Code | claude.ai/code |
curl | Pre-installed on macOS, Linux, Windows 10+ |
/plugin marketplace add https://github.com/adorosario/customgpt-skill-plugin
/plugin install customgpt-ai-rag
/reload-plugins
/plugin marketplace update https://github.com/adorosario/customgpt-skill-plugin
/plugin install customgpt-ai-rag
/reload-plugins
The plugin resolves your API key automatically, checking these locations in order:
$CUSTOMGPT_API_KEY.env file — walks up from $PWD looking for CUSTOMGPT_API_KEY=~/.claude/customgpt-config.json → apiKey fieldTo save your key for all projects:
echo '{"apiKey":"YOUR_KEY_HERE"}' > ~/.claude/customgpt-config.json
If no key is found, the plugin will prompt you and save it automatically.
1. /create-agent → index this project into a new CustomGPT.ai agent
2. /check-status → wait until indexing completes (index_status: ok)
3. /query-agent → ask questions across your entire codebase
you: /create-agent
→ uploads all eligible files in the project to a new CustomGPT.ai agent
→ saves agent ID to .customgpt-meta.json
you: /check-status
→ shows indexing progress: pages_found, pages_indexed, index_status
you: /query-agent where is authentication handled?
→ returns an AI answer with citations pointing to the relevant files
| Skill | Argument | What it does |
|---|---|---|
/create-agent | [folder path] | Create a new agent and upload an entire folder. Defaults to the Git root or $PWD. |
/index-files | [file, directory, or blank] | Add specific files or a directory to an existing agent. Auto-creates an agent if none exists. |
/refresh-agent | — | Wipe all indexed documents and re-upload everything. Use after large changes. |
/reindex-file | [filename or path] | Delete and re-upload a single changed file. Handles both uploaded files and URL-based documents. |
| Skill | Argument | What it does |
|---|---|---|
/query-agent | [your question] | Ask a plain-language question. Returns an AI-generated answer with source citations. Warns if files have changed since the last index. |
| Skill | Argument | What it does |
|---|---|---|
/check-status | — | Agent metadata, chat availability, and indexing statistics. |
/list-pages | — | All indexed documents with their IDs and crawl/index status. |
/check-page | [filename or URL] | Look up whether a specific file is indexed. |
/delete-page | [page ID] | Remove a document from the knowledge base by page ID. |
/delete-agent | — | Permanently delete the agent and remove .customgpt-meta.json. Requires confirmation. |
| Skill | Argument | What it does |
|---|---|---|
/help | [setup | index | query | manage] | Show available skills, prerequisites, and routing guidance. |
You don't have to type slash commands. The plugin activates on natural language too:
"index this repo" → /create-agent
"add this file" → /index-files
"search my codebase" → /query-agent
"what does my code say about X" → /query-agent
"show all indexed files" → /list-pages
"re-sync everything" → /refresh-agent
Code: .js .ts .tsx .jsx .py .go .rb .java .cs .cpp .rs .swift .kt .php .sh .html .css .scss .vue .svelte .sql .graphql .proto and more
Config / Data: .json .yaml .toml .xml .ini .cfg .env.example
Docs: .md .rst .txt .pdf .docx .pptx .xlsx .csv
npx claudepluginhub adorosario/customgpt-skill-plugin --plugin customgpt-ai-ragUse probe to search indexed project knowledge from docs and code through MCP.
Optimized file search, semantic indexing, and persistent memory for Claude Code — with optional sync to a self-hosted web dashboard
Local RAG system with embedded Multi-Agent Framework for Claude Code plugin
Document search with hybrid BM25/semantic retrieval, GraphRAG knowledge graphs, and pluggable providers for Claude Code. Index documentation and code, then search using keyword matching, semantic similarity, graph relationships, or comprehensive multi-mode fusion.
Local-first memory server — hybrid BM25+vector search, vault management, lint, and launchd lifecycle for project knowledge.
Full AI context layer over MCP — tree-sitter code-map, document RAG (PDF/Office/HTML/email + OCR + reranker), shared agent memory, on-demand web crawl, git history + blame + per-symbol diff. 300+ languages, 8 coding-agent harnesses, content-addressed Fjall + LanceDB.