Parseable

io.github.parseablehqPublished by GitHub account parseablehqv0.2.16Updated Oct 24, 2026

Model Context Protocol server for Parseable. Lets LLMs discover and query Parseable.

Overview

AI-generated overview

Connects an MCP client to a Parseable instance so the assistant can discover, query, and manage datasets and alerts in natural language.

What you can do
An adapter that forwards MCP tool discovery and calls to a connected Parseable instance. The tool catalog is fetched from the instance's LLM tools endpoint, so OSS, Enterprise, and Cloud deployments can expose different tool sets. Queries are executed server-side with argument validation and RBAC, and the README notes that query_sql blocks DDL/DML and enforces a row limit and a mandatory time window, while evaluate_alert can fire real notifications.
When to use it
Use it when you already run or subscribe to Parseable and want an assistant to explore logs or event datasets, run SQL-style queries, and inspect or manage alerts without leaving the chat client. It is not useful without a reachable Parseable instance and an API key.
Requirements
Runs locally over stdio via the npm package, or as a self-hosted HTTP process. Stdio mode needs PARSEABLE_URL and PARSEABLE_API_KEY; HTTP mode needs PORT and, for cloud, PARSEABLE_ORCHESTRATOR_URL and PARSEABLE_CLOUD_AUTH_TOKEN, with clients sending X-Parseable-URL and X-API-Key headers. Node.js is required for the npm package; optional OpenTelemetry variables exist.
Before you install
The API key is stored in MCP client configuration, so use a key scoped to minimum permissions and HTTPS for remote deployments. evaluate_alert can send real notifications, and query_sql is restricted to read-only queries with a row limit and mandatory time window. Private and loopback Parseable URLs are rejected by default; PARSEABLE_MCP_ALLOW_PRIVATE should only be enabled for trusted deployments. The README states there is no telemetry and outbound calls go only to the configured Parseable…

Installation

In SourceWeft

  1. Open Parseable in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

Parseable MCP Server

Model Context Protocol server for Parseable. Lets any MCP-capable client (Claude Desktop, Claude Code, Cursor, VS Code Copilot, Windsurf, Continue, Cline, Zed, Codex) discover, query, and manage Parseable datasets and alerts using natural language.

Two transports:

ModeTransportAuthUse when
stdioStdin/stdoutAPI key via env varsClaude Desktop, Cursor, VS Code, local clients
httpStreamable HTTPCloud API key, or self-hosted URL + API keyHosted deployments and remote clients

Quickstart — stdio (local)

One command — interactive setup, detects Claude Desktop / Cursor, writes config files:

bash
npx -y @parseable/parseable-mcp-server init

Restart your MCP client. Tools appear. Done.

Scripted:

bash
npx -y @parseable/parseable-mcp-server init \  --client claude-desktop \  --url https://your-parseable.example.com \  --api-key "$PARSEABLE_API_KEY"

Supported --client values: claude-desktop, cursor.

Cursor Marketplace plugin

The repository also contains a native Cursor plugin manifest. Marketplace installs connect to https://mcp.parseable.com/mcp and read credentials from the environment.

Set these variables before starting Cursor:

bash
export PARSEABLE_URL="https://your-parseable.example.com"export PARSEABLE_API_KEY="your-api-key"export PARSEABLE_MODE="self-hosted"

PARSEABLE_API_KEY and PARSEABLE_MODE are required by the Cursor plugin. Use PARSEABLE_MODE=cloud and omit PARSEABLE_URL for Parseable Cloud. Use PARSEABLE_MODE=self-hosted with PARSEABLE_URL for self-hosted Parseable.

bash
# Parseable Cloudexport PARSEABLE_MODE="cloud"export PARSEABLE_API_KEY="your-cloud-api-key"unset PARSEABLE_URL

Then install the Parseable plugin from the Cursor Marketplace. If Cursor was opened from the macOS dock or another GUI launcher that does not inherit shell variables, use the interactive setup instead:

bash
npx -y @parseable/parseable-mcp-server init --client cursor

Plugin metadata lives in .cursor-plugin/plugin.json; its MCP connection configuration lives in mcp.json. Keep the plugin version aligned with the npm package version when releasing.


Quickstart — HTTP (hosted)

HTTP mode serves a setup page and MCP endpoint from one process. Cloud clients supply an API key. Self-hosted clients supply their Parseable URL and API key.

1. Set env vars

bash
# .envPORT=8787

2. Run

bash
# From sourcenpm run build:allnode dist/server.js http
# Dockerdocker build -t parseable-mcp-server .docker run -p 8787:8787 --env-file .env parseable-mcp-server

3. Connect from Claude

  • Claude Desktop → Settings → Connectors → Add custom connector
  • Name: Parseable
  • URL: https://mcp.your-domain.com/mcp
  • Header X-Parseable-URL: https://your-parseable.example.com
  • Header X-API-Key: your Parseable API key
  • Click Add → Connect

4. Connect from Claude Code

bash
claude mcp add --transport http parseable https://mcp.your-domain.com/mcp --scope user \  --header "X-Parseable-URL: https://your-parseable.example.com" \  --header "X-API-Key: $PARSEABLE_API_KEY"

5. Connect from Cursor / VS Code

json
{  "mcpServers": {    "parseable": {      "type": "http",      "url": "https://mcp.your-domain.com/mcp",      "headers": {        "X-Parseable-URL": "https://your-parseable.example.com",        "X-API-Key": "your-parseable-api-key"      }    }  }}

HTTP authentication

Each POST /mcp request requires X-API-Key and supports two modes:

ModeHeaders
CloudX-Parseable-Mode: cloud, X-API-Key
Self-hosted (default)X-Parseable-URL, X-API-Key

X-Parseable-Mode is checked first when present. Omitting it selects self-hosted mode; clients do not need to send X-Parseable-Mode: self-hosted. In cloud mode, server validates API key with Parseable Cloud, caches returned URL and tenant routing in a bounded in-memory LRU for 24 hours, and sends x-p-tenant on Parseable requests. Cache is disposable; misses and process restarts resolve through Cloud again.

For self-hosted mode, HTTP server validates supplied URL and forwards API key to that Parseable instance. By default, private and loopback Parseable URLs are rejected to limit SSRF. Set PARSEABLE_MCP_ALLOW_PRIVATE=true only for trusted deployments that need private network targets.


Environment variables

stdio mode

VarRequiredDefaultPurpose
PARSEABLE_URL✅—Parseable base URL
PARSEABLE_API_KEY✅—API key for self-hosted Parseable
PARSEABLE_QUERY_TIMEOUT_MS30000HTTP timeout (ms)

HTTP mode

VarRequiredDefaultPurpose
PORT8787HTTP listen port
PARSEABLE_MCP_ALLOW_PRIVATEfalsePermit private/loopback Parseable URLs supplied in request headers
PARSEABLE_ORCHESTRATOR_URLCloud only-Parseable Cloud orchestrator base URL
PARSEABLE_CLOUD_AUTH_TOKENCloud only-Service bearer token for API-key validation
PARSEABLE_CLOUD_CACHE_TTL_SECONDS86400Cloud routing LRU TTL
PARSEABLE_CLOUD_CACHE_MAX_ENTRIES10000Maximum cached cloud API-key routes
PARSEABLE_CLOUD_VALIDATE_TIMEOUT_MS10000Cloud validation timeout

OpenTelemetry (optional)

VarDefaultPurpose
PARSEABLE_OTEL_ENABLEDfalseEnable trace export to Parseable
PARSEABLE_OTEL_ENDPOINT—Parseable OTLP endpoint
PARSEABLE_OTEL_USERNAME—Basic auth for OTLP
PARSEABLE_OTEL_PASSWORD—Basic auth for OTLP
PARSEABLE_OTEL_TRACES_STREAMmcp-tracesStream name for traces
PARSEABLE_OTEL_DEBUGfalseLog OTLP export errors

Copy .env.example → .env for a full template.


Tools

Tools are owned by the connected Parseable instance, not this adapter. For MCP tools/list, the adapter fetches the edition-specific catalog from GET /api/prism/v1/llm/tools/list. OSS, Enterprise, and Cloud can expose different tool sets.

The adapter caches the catalog for its server-provided max-age and revalidates it with ETag. MCP tools/call requests are forwarded to POST /api/prism/v1/llm/tools/call, where Parseable validates arguments, applies RBAC, executes the tool, and returns an MCP CallToolResult.


Client setup — stdio

Claude Desktop

~/Library/Application Support/Claude/claude_desktop_config.json

json
{  "mcpServers": {    "Parseable": {      "command": "npx",      "args": ["-y", "@parseable/parseable-mcp-server"],      "env": {        "PARSEABLE_URL": "https://your-parseable.example.com",        "PARSEABLE_API_KEY": "your-api-key"      }    }  }}

Claude Code

bash
claude mcp add Parseable \  --env PARSEABLE_URL=https://your-parseable.example.com \  --env PARSEABLE_API_KEY=your-api-key \  -- npx -y @parseable/parseable-mcp-server

Cursor

~/.cursor/mcp.json:

json
{  "mcpServers": {    "Parseable": {      "command": "npx",      "args": ["-y", "@parseable/parseable-mcp-server"],      "env": {        "PARSEABLE_URL": "https://your-parseable.example.com",        "PARSEABLE_API_KEY": "your-api-key"      }    }  }}

VS Code

.vscode/mcp.json:

json
{  "servers": {    "Parseable": {      "type": "stdio",      "command": "npx",      "args": ["-y", "@parseable/parseable-mcp-server"],      "env": {        "PARSEABLE_URL": "https://your-parseable.example.com",        "PARSEABLE_API_KEY": "your-api-key"      }    }  }}

Development

bash
git clone https://github.com/parseablehq/parseable-mcp-server.gitcd parseable-mcp-servernpm installcp .env.example .env    # fill in your values
# Buildnpm run build           # server only (tsc)npm run build:ui        # React UI only (vite)npm run build:all       # both
# Runnode dist/server.js         # stdio modenode dist/server.js http    # HTTP mode (port 8787)
# Devnpm run dev             # tsc --watchnpm run dev:ui          # vite dev server (proxies API to :8787)npm testnpm run lintnpm run fix             # biome auto-fix

CI (GitHub Actions) runs lint + build:all + test on every push/PR to main on Node 22. On merge to main, Docker image is published to ghcr.io/parseablehq/parseable-mcp-server.


Security

  • Parseable API keys live in MCP client configuration. Use keys scoped to minimum required permissions.
  • HTTP clients send credentials in X-Parseable-URL and X-API-Key; always use HTTPS for remote deployments.
  • query_sql blocks DDL/DML and enforces a row LIMIT. Time window is mandatory.
  • evaluate_alert can fire real notifications — review the call before approving.
  • No telemetry. Outbound calls go only to the Parseable instance configured by the user.

License

Apache-2.0. See LICENSE.

Source: README.md at commit a65d863

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.2.16LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts