probe-search 0.4.2

AI Agent Context Engine — give your coding agent a brain beyond code

probe

Give your AI agent a brain beyond code.

probe is a CLI tool and MCP server that indexes your project's docs and code, then serves curated, reranked context to AI coding agents in milliseconds.

---

Contents

• Why probe
• Quick Start
• MCP Server Setup
• Indexing
• How It Works
• Example Output
• CLI Reference
• Configuration
• Data Handling

---

Why probe

AI coding agents often answer project questions by grepping, reading files one by one, and assembling context over several tool calls. probe gives them one ranked search across docs and code.

What you get:

• Claude Code plugin with bundled MCP server and usage skill
• Manual Claude Code setup with probe install
• First-use indexing and incremental refresh-before-search
• Hybrid keyword + semantic retrieval across docs, code, text, and PDFs
• Cross-encoder reranking with ZeroEntropy zerank-2
• Local index storage in .probe/
• MCP tools and resources for Claude Code, Cursor, and other MCP-compatible agents

---

Quick Start

Claude Code plugin (recommended)

Get a free ZeroEntropy API key at https://dashboard.zeroentropy.dev. Then run inside Claude Code:

/plugin marketplace add https://github.com/zeroentropy-ai/probe.git
/plugin install probe@zeroentropy
/reload-plugins

Use the HTTPS URL above. The zeroentropy-ai/probe shorthand makes Claude Code clone over SSH, which requires a configured GitHub SSH key. The Claude Code slash command treats --sparse as part of the URL, so do not pass sparse checkout options there.

Claude Code asks for your ZeroEntropy API key during plugin install. The plugin starts probe with uvx --from probe-search==0.4.2 probe mcp, so Claude Code does not need a separate probe install.

If you use the claude plugin install shell command instead of the /plugin slash command, export ZEROENTROPY_API_KEY before starting Claude Code. From a shell, sparse checkout options are supported:

claude plugin marketplace add https://github.com/zeroentropy-ai/probe.git --sparse .claude-plugin plugins
claude plugin install probe@zeroentropy

Manual CLI/MCP setup

uv tool install probe-search
# or: pipx install probe-search

export ZEROENTROPY_API_KEY="ze_xxx"
probe install
claude mcp list

Use uv tool install or pipx so the registered probe path stays valid. If you install probe in a temporary virtual environment, rerun probe install after recreating that environment.

For CLI-only use:

pip install probe-search
export ZEROENTROPY_API_KEY="ze_xxx"
probe index .
probe search "how does authentication work"

Verify your setup

probe doctor
probe smoke

probe doctor checks your API key, Claude Code wiring, MCP registration, and local index health without printing secrets or uploading project content. probe smoke indexes a tiny sample project and confirms search works. Use probe smoke --current to validate the current repo.

---

MCP Server Setup (Claude Code, Cursor)

The Claude Code plugin is the best default. For Cursor or advanced use, add a .mcp.json file to your project root:

{
  "mcpServers": {
    "probe": {
      "command": "uvx",
      "args": ["--from", "probe-search", "probe", "mcp"],
      "env": {
        "ZEROENTROPY_API_KEY": "ze_xxx"
      }
    }
  }
}

Your agent gets four tools:

Tool	What it does
probe_search	Search docs and code with automatic refresh and reranking
probe_index	Index or re-index project files
probe_status	Show what's indexed
probe_read	Read a file, optionally with focused line ranges

It also gets MCP resources:

Resource	What it exposes
probe://status	Index status and provider configuration
probe://files	Indexed file list
probe://file/{path}	Read a project file by URL-encoded path

When Claude Code starts probe, probe uses CLAUDE_PROJECT_DIR as the project root. That keeps search, indexing, and file reads pinned to the repo even when the MCP process starts from another directory.

---

Indexing

probe indexes Markdown, MDX, plain text, reStructuredText, AsciiDoc, TeX, YAML, JSON, PDFs, and code in Python, JavaScript, TypeScript, TSX, JSX, Go, Rust, and Java.

File discovery respects root .gitignore and .probeignore files. It also skips .git/, .probe/, __pycache__/, .venv/, and *.pyc.

On first search, probe creates the local .probe/ index automatically. Before later CLI or MCP searches, it checks for added, changed, and deleted files, then refreshes only affected chunks. Set PROBE_REFRESH_TTL=0 to check before every search, or PROBE_REFRESH_TTL=-1 to disable automatic refresh.

Chunks keep Markdown header paths, code symbol names, PDF page numbers, and line ranges so agents can cite focused source locations.

---

How It Works

1. Hybrid retrieval: each query uses semantic vector search and SQLite FTS5 keyword search.
2. Reranking: candidates are fused and reranked with ZeroEntropy zerank-2.
3. Context assembly: results are deduplicated, trimmed to your token budget, and returned with file, section, and line metadata.

---

Example Output

$ probe search "how does authentication work"

 Found 5 results (342 chunks searched)

 [0.94] docs/design/auth.md > Authentication > OAuth Flow
   We use PKCE-based OAuth 2.0 with Auth0 as the identity provider.
   The flow works as follows: 1) Client generates a code verifier...

 [0.87] src/auth/oauth.py:42-71 > class OAuthHandler
   class OAuthHandler:
       """Handles OAuth2 PKCE flow for web and mobile clients."""
       def __init__(self, client_id: str, redirect_uri: str):

 ------------------------------------------
 zembed-1 + zerank-2 | 1,847 tokens | 0.3s

For scripts and agents:

probe search "how does authentication work" --json
probe status --json
probe doctor --json
probe smoke --json

---

CLI Reference

Command	Description
probe index [paths...]	Index project files for semantic search
probe index --full	Force full re-index
probe install	Register probe as a user-scope MCP server in Claude Code
probe search "query"	Search project knowledge with natural language
probe search --top-k N	Limit number of results
probe search --type code	Filter by file type
probe search --no-rerank	Skip reranking
probe search --max-tokens N	Set result token budget
probe search --json	Emit machine-readable results with line ranges
probe status	Show index stats and model config
probe status --json	Emit machine-readable index status
probe list	List indexed files
probe config	Show current model configuration
probe init	Create local config from environment
probe doctor	Diagnose API key, Claude Code, MCP, and index setup
probe doctor --json	Emit machine-readable diagnostics
probe smoke	Run an end-to-end indexing and search validation
probe smoke --current	Smoke-test the current project
probe mcp	Start MCP server
probe uninstall [--purge]	Unregister probe; --purge also deletes .probe/

---

Configuration

probe stores its index and config in .probe/ at your project root. Add .probe/ to .gitignore.

# .probe/config.yaml
providers:
  embedding:
    name: zeroentropy
    model: zembed-1
    dimensions: 1280
  reranker:
    name: zeroentropy
    model: zerank-2

---

Data Handling

Documents are chunked and stored locally in .probe/ with SQLite and numpy. Only chunk text is sent to the embedding and reranking API for processing. Documents are never uploaded or stored on an external server.

---

Links

• ZeroEntropy -- the AI retrieval engine powering probe
• API Dashboard -- get your API key
• Documentation -- ZeroEntropy API docs
• Models -- zembed-1 and zerank-2 details

---

License

MIT -- see LICENSE for details.