> For the complete documentation index, see [llms.txt](https://syv-labs.gitbook.io/context-pool/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://syv-labs.gitbook.io/context-pool/reference/cli.md).

# CLI Reference

## `cxp install [OPTIONS]`

Register the contextpool MCP server with Claude Code and Cursor, and configure an LLM backend.

```bash
cxp install                          # register MCP + run backend wizard
cxp install --setup                  # re-run just the backend wizard
cxp install --force                  # overwrite existing MCP entries
cxp install --skip-claude            # skip Claude Code registration
cxp install --skip-cursor            # skip Cursor registration
cxp install --skip-setup             # skip the backend wizard (non-interactive installs)
cxp install --binary-path <path>     # path to register (defaults to current executable)
cxp install --claude-json <path>     # override ~/.claude.json path
cxp install --cursor-mcp <path>      # override ~/.cursor/mcp.json path
```

The wizard presents four backends as distinct options:

| Choice        | Description                                                                  |
| ------------- | ---------------------------------------------------------------------------- |
| Claude Code   | Free, uses your existing Claude Code subscription. Spawns `claude -p`.       |
| Anthropic API | Direct HTTP to the Messages API. Fastest, parallelizes well, works headless. |
| OpenAI API    | OpenAI chat completions. Good if you already have a key.                     |
| NVIDIA NIM    | OpenAI-compatible endpoint via NVIDIA.                                       |

Keys are saved to the system keychain and a `0600` local file — no env vars required after setup.

***

## `cxp init`

Extract and store insights from your IDE sessions into the current project.

### `cxp init claude-code [session-ids...] [OPTIONS]`

```bash
cxp init claude-code                       # all sessions for this project
cxp init claude-code <id1> <id2>          # specific session ids
cxp init claude-code --local               # write to ./ContextPool/
cxp init claude-code --out ./store        # custom output directory
cxp init claude-code --claude-dir <path>  # custom ~/.claude root
```

### `cxp init cursor [chat-ids...] [OPTIONS]`

```bash
cxp init cursor                            # all Cursor chats for this project
cxp init cursor <id1> <id2>               # specific chat ids
cxp init cursor --local                    # write to ./ContextPool/
cxp init cursor --out ./store             # custom output directory
cxp init cursor --cursor-dir <path>       # custom ~/.cursor root
```

Sessions with no extractable insights are skipped — no empty `.summary.md` files are written.

***

## `cxp export`

Bulk-export and summarize transcripts from any supported IDE.

### `cxp export claude-code [OPTIONS]`

```bash
cxp export claude-code                     # all Claude Code sessions
cxp export claude-code --session <path>   # single .jsonl session file
cxp export claude-code --out ./out        # custom output directory
cxp export claude-code --claude-dir <path>
cxp export claude-code --offline          # skip LLM, write fallback summaries
```

### `cxp export cursor [OPTIONS]`

```bash
cxp export cursor                          # all Cursor transcripts
cxp export cursor --transcript <path>     # single .jsonl file
cxp export cursor --out ./out             # custom output directory
cxp export cursor --cursor-dir <path>
cxp export cursor --offline
```

### `cxp export vscdb [OPTIONS]`

Export from VS Code-style `state.vscdb` workspace storage (Cursor, Windsurf, other forks).

```bash
cxp export vscdb --product Cursor
cxp export vscdb --product Windsurf
cxp export vscdb --product Cursor --workspace-storage "<custom-path>"
```

Default paths:

* macOS (Cursor): `~/Library/Application Support/Cursor/User/workspaceStorage`
* Windows (Cursor): `%APPDATA%\Cursor\User\workspaceStorage`

### `cxp export kiro [OPTIONS]`

Export a Kiro session exported with `/chat save <path>`.

```bash
cxp export kiro --chat-json ./kiro-session.json
```

***

## `cxp mcp [OPTIONS]`

Start the MCP server. Used by your IDE config — not normally called directly.

```bash
cxp mcp
cxp mcp --data-dir ./ContextPool   # custom data directory
```

***

## `cxp auth [team-key] [OPTIONS]`

Authenticate with a ContextPool team API key.

```bash
cxp auth cxp_team_4c4143d879ce471ab14afacb081a7b4f   # save key to keychain
cxp auth --status                                      # show current auth status
cxp auth --logout                                      # remove stored API key
```

***

## `cxp push [OPTIONS]`

Push local insights to the team cloud pool.

```bash
cxp push
cxp push --all      # push all local projects, not just current directory
cxp push --dry-run  # show what would be pushed without pushing
```

***

## `cxp pull [OPTIONS]`

Pull team insights to local cache.

```bash
cxp pull
cxp pull --all      # pull all team projects, not just current directory
```

***

## `cxp team [ACTION]`

Show team info or list projects.

```bash
cxp team              # show team info
cxp team projects     # list team projects with insight counts
```

***

## Global Flags

```bash
cxp --reset-nvidia-api-key    # clear saved NVIDIA key from keychain
cxp --version                 # print version
cxp --help                    # list commands
```

To reset Anthropic or OpenAI keys, re-run `cxp install --setup` and choose a new backend.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://syv-labs.gitbook.io/context-pool/reference/cli.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
