Skip to main content
Glama

simpletel-mcp

Model Context Protocol server for simpletel — ask your coding agent "what just happened?" and let it read the OpenTelemetry traces, logs, metrics, and account usage of your running services.

simpletel mcp is a stdio MCP server built into the simpletel CLI. It speaks newline-delimited JSON-RPC 2.0 on stdin/stdout (MCP protocol revision 2025-06-18) and exposes simpletel's read commands as native agent tools. It is a single static binary: no runtime, no dependencies, no background service.

  • One-line install of the CLI: curl -fsSL https://simpletel.dev/install | sh

  • Run it as simpletel mcp

  • Or use the container: ghcr.io/willnewby/simpletel-mcp

  • Registry name: io.github.willnewby/simpletel


Tools

The server advertises four tools. Every tool returns text (the JSON document the matching CLI command produces), and tool-side failures come back as a normal result with isError: true — a tool call never breaks the session.

Tool

Arguments

What it does

get_traces

since (default 1h), service, errors_only, min_duration, limit

Lists this team's traces newest-first. The JSON document simpletel get traces --format json produces. Use it to find a failing or slow request.

get_trace

trace_id (required)

Fetches one trace in full — spans, span events, and correlated log records. The JSON document simpletel get trace <id> --format json produces.

verify

service (required), timeout (default 60s)

Runs the simpletel verify onboarding check: polls until service has a fresh trace carrying spans from a library instrumentation. Returns the rung report and the exit status (0 verified, 1 deadline miss). A deadline miss is a completed check, not a tool error.

usage

(none)

Shows the account's plan, month-to-date records, allowance, and spend cap. The JSON document simpletel usage --format json produces. Account-scoped (needs a logged-in user token).

get_traces, get_trace, and verify are team-scoped; usage is account-scoped.


Related MCP server: @originchain/telemetry-mcp

Authentication and team resolution

The server resolves one connection at startup, using the same precedence as every other simpletel read command: **flag > environment variable > config file

default**.

Setting

Flag

Environment

Config file

Default

Endpoint

--endpoint

SIMPLETEL_ENDPOINT

endpoint

http://localhost:4318

Team

--team

SIMPLETEL_TEAM

current_team

(none)

Admin token

--admin-token

SIMPLETEL_ADMIN_TOKEN

—

(none)

User token (for usage)

—

SIMPLETEL_TOKEN

auth token

(none)

Config path

--config

SIMPLETEL_CONFIG

—

~/.simpletel/config.toml

  • The config file is ~/.simpletel/config.toml, written by simpletel login and simpletel team (mode 0600). simpletel login signs you in with GitHub; simpletel team create / simpletel team use set the current team.

  • --team accepts either a 20-hex team id — used verbatim, no token needed — or a friendly name. A name is resolved only, never created, and needs an admin token to look it up.

  • usage additionally needs a user token: run simpletel login, or set SIMPLETEL_TOKEN (a stk_… token).

If the team is missing or wrong the server still answers initialize and tools/list, so the client sees the tool catalog; each team-scoped call then reports the problem as its own isError result.

Nothing is ever sent to simpletel by this server beyond the read requests the tools make to the endpoint you configure. Diagnostics go to stderr; stdout carries protocol messages only.


Install

1. Install the CLI

curl -fsSL https://simpletel.dev/install | sh

Installs simpletel into ~/.simpletel/bin/ (verify the sha256; no sudo, never prompts). Then sign in and create/select a team:

simpletel login            # GitHub device flow
simpletel team create --name my-team

2. Register the server with your client

The MCP subcommand is simpletel mcp. It takes the connection flags --endpoint, --team, --admin-token, and --config.

Claude Code

# by 20-hex team id (no extra credentials needed)
claude mcp add simpletel -- ~/.simpletel/bin/simpletel mcp --team <20-hex-team-id>

# or rely on the logged-in config file (~/.simpletel/config.toml)
claude mcp add simpletel -- ~/.simpletel/bin/simpletel mcp

Pass connection values as environment instead with --env:

claude mcp add simpletel \
  --env SIMPLETEL_ENDPOINT=https://simpletel.dev \
  --env SIMPLETEL_TEAM=<20-hex-team-id> \
  -- ~/.simpletel/bin/simpletel mcp

Claude Desktop / any JSON-configured client (claude_desktop_config.json, Cursor, Cline, Windsurf, …):

{
  "mcpServers": {
    "simpletel": {
      "command": "/Users/you/.simpletel/bin/simpletel",
      "args": ["mcp"],
      "env": {
        "SIMPLETEL_ENDPOINT": "https://simpletel.dev",
        "SIMPLETEL_TEAM": "<20-hex-team-id>"
      }
    }
  }
}

If you are logged in locally you can drop env and the command will read ~/.simpletel/config.toml. In a sandboxed client that cannot read your home directory, pass the values through env (or --env / --router equivalents) instead.


Docker

The image bundles the CLI and runs simpletel mcp. It is built for linux/amd64 and linux/arm64:

docker run -i --rm ghcr.io/willnewby/simpletel-mcp:latest

The container starts with no credentials. Provide them either by mounting the config file or through environment variables (recommended — nothing is written to the image).

Mount an existing config file (read-only). The image runs as the distroless nonroot user (uid/gid 65532) with HOME=/home/nonroot, so the CLI's default config path inside the container is /home/nonroot/.simpletel/config.toml. Because simpletel login writes that file mode 0600 for you, run the container as the file's owner so it stays readable:

docker run -i --rm \
  --user "$(id -u):$(id -g)" \
  -v "$HOME/.simpletel/config.toml:/home/nonroot/.simpletel/config.toml:ro" \
  ghcr.io/willnewby/simpletel-mcp:latest

(Passing the connection through -e below avoids the ownership question entirely and is the recommended path.)

Or pass everything through the environment:

docker run -i --rm \
  -e SIMPLETEL_ENDPOINT=https://simpletel.dev \
  -e SIMPLETEL_TEAM=<20-hex-team-id> \
  ghcr.io/willnewby/simpletel-mcp:latest

As a JSON client entry:

{
  "mcpServers": {
    "simpletel": {
      "command": "docker",
      "args": [
        "run", "-i", "--rm",
        "-e", "SIMPLETEL_ENDPOINT=https://simpletel.dev",
        "-e", "SIMPLETEL_TEAM=<20-hex-team-id>",
        "ghcr.io/willnewby/simpletel-mcp:latest"
      ]
    }
  }
}

The MCP transport is stdio, so -i (keep stdin open) is required and you must not add -t.


Configuration reference

simpletel mcp [--endpoint URL] [--team <id-or-name>] [--admin-token TOKEN] [--config PATH]

Variable

Purpose

SIMPLETEL_ENDPOINT

Server base URL — no /v1/traces suffix (default http://localhost:4318).

SIMPLETEL_TEAM

20-hex team id, or a friendly name (name needs an admin token).

SIMPLETEL_ADMIN_TOKEN

Token used to resolve a team name; not needed for a 20-hex id.

SIMPLETEL_TOKEN

User token (stk_…) used by the usage tool.

SIMPLETEL_CONFIG

Config file path (default ~/.simpletel/config.toml).


Limitations

  • stdio only. simpletel mcp is launched as a subprocess by the client; it is not an HTTP/SSE endpoint.

  • Read-only. The tools query traces, usage, and onboarding state. They do not ingest data or change account settings.

  • The host CLI is required unless you use the container. The simpletel binary is fetched at build time by the Dockerfile; the repo ships no binary of its own — see Dockerfile for the build.

  • Windows is not a first-class target of the one-line installer; use the Docker image or WSL.


Repository contents

File

Purpose

Dockerfile

Builds the stdio image from the published CLI.

server.json

Official MCP Registry manifest (OCI package). CI rewrites the version/tag at publish time.

glama.json

Maintainer metadata for Glama auto-indexing.

llms-install.md

Install instructions for LLM agents / directory crawlers.

.github/workflows/publish.yml

Builds and pushes ghcr.io/willnewby/simpletel-mcp and publishes to the registry.

License

MIT — see LICENSE.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables coding agents to access live runtime observability data such as logs, deploys, and health metrics for evidence-based incident triage.
      1
      MIT
    • A
      license
      A
      quality
      C
      maintenance
      Enables AI agents to query OriginChain's telemetry agent for log patterns, semantic incident analysis, distributed traces, service graphs, and read-only SQL queries over telemetry data.
      10
      4 npm
      MIT
    • F
      license
      Not graded
      quality
      A
      maintenance
      Enables AI agents and MCP clients to debug and observe LLM applications by querying traces, sessions, exceptions, prompts, datasets, scores, and metrics. Supports local stdio/HTTP access, selective tool loading, and read-only operation for safer observability workflows.
      -