Skip to main content

​
Two paths

Pick the path that fits your situation:
PathTimeBest for
Local / dev60 secondsChecking your own Claude/Cursor/VS Code setup
Production5 minutesMonitoring remote MCP servers in a deployed stack

​
Local / dev

You already use Claude Desktop, Cursor, or another MCP-enabled IDE. LangSight reads your existing config files and starts monitoring immediately.

​
Step 1 — Install

pip install langsight
# or with uv:
uv tool install langsight

​
Step 2 — Discover your servers

langsight init
LangSight scans every MCP config file on your machine — Claude Desktop, Cursor, VS Code, Windsurf, Claude Code, Gemini CLI, Kiro, Zed, Cline, and any project-local configs. It writes a .langsight.yaml and immediately runs a first health check so you see results right away. 
LangSight Init — scanning for MCP servers...

  Claude Desktop  (~/.config/claude/claude_desktop_config.json)  4 servers
  Cursor          (~/.cursor/mcp.json)                           2 servers
  VS Code         (~/.vscode/mcp.json)                           1 server

Discovered 7 MCP servers:
  1. snowflake-mcp    (stdio)
  2. github-mcp       (stdio)
  3. slack-mcp        (sse)
  4. jira-mcp         (stdio)
  5. postgres-mcp     (stdio)
  6. filesystem-mcp   (stdio)
  7. search-mcp       (stdio)

Include all 7 servers? [Y/n]: Y

Running first health check...

  snowflake-mcp    ✓ UP       89ms
  github-mcp       ✓ UP       54ms
  slack-mcp        ✓ UP      142ms
  jira-mcp         ✗ DOWN    —       timeout after 5s
  postgres-mcp     ✓ UP       31ms
  filesystem-mcp   ✓ UP       12ms
  search-mcp       ✓ UP       67ms

6/7 servers healthy
  jira-mcp is DOWN — check the process is running

Config written to .langsight.yaml

​
Step 3 — Health table

langsight mcp-health
Connects to every server, checks latency, detects schema drift, and shows a live table.

​
Step 4 — Scorecard

langsight scorecard

MCP Server Scorecards
──────────────────────────────────────────────────────────────────
Server           Grade   Score   Avail   Security   Reliability
snowflake-mcp      A      92      30/30    23/25       19/20
github-mcp         A      90      30/30    22/25       18/20
slack-mcp          B      84      29/30    20/25       17/20
jira-mcp           F      —       —        —            —       consecutive failures (10+)
postgres-mcp       A+     97      30/30    25/25       20/20
filesystem-mcp     A      91      30/30    22/25       19/20
search-mcp         B      81      28/30    21/25       16/20

​
Production

You have MCP servers running as HTTP services (behind a load balancer, in Kubernetes, etc.). Define them in a YAML config and run LangSight via Docker.

​
Step 1 — Write a config

# .langsight.yaml
servers:
  - name: postgres-mcp
    transport: streamable_http
    url: https://postgres-mcp.internal.company.com/mcp

  - name: search-mcp
    transport: streamable_http
    url: https://search-mcp.k8s.company.com/mcp
    headers:
      Authorization: "Bearer ${SEARCH_MCP_TOKEN}"

  - name: github-mcp
    transport: streamable_http
    url: https://github-mcp.prod.company.com/mcp
    headers:
      X-API-Key: "${GITHUB_MCP_KEY}"

alerts:
  slack_webhook: "${SLACK_WEBHOOK_URL}"
Environment variables in ${VAR} syntax are expanded at runtime — never put secrets in the YAML itself.

​
Step 2 — Start the stack

docker compose up langsight
LangSight starts health-checking immediately. Open http://localhost:3003 to see the dashboard (Docker only — without Docker: dashboard at localhost:3002, API at localhost:8000).

​
Step 3 — Add servers later

You don’t have to restart Docker to add a server. Use langsight add: 
langsight add billing-mcp --url https://billing-mcp.internal.company.com/mcp \
  --header "Authorization=Bearer $BILLING_TOKEN"
See Server Discovery for the full langsight add reference.

​
What’s next

Server Discovery

All 10+ IDE clients LangSight discovers. How to add servers manually with langsight add.

Health Monitoring

Health check protocol, status states, continuous monitoring daemon.

Schema Drift

Detect breaking and compatible tool schema changes before they cause failures.

Server Scorecard

A-F composite grade across availability, security, reliability, and performance.

​
Project scoping (optional)

The steps above work with no project configuration. When you are ready to integrate with the LangSight dashboard and share monitoring data with your team, you can scope all CLI output to a specific project using a project-scoped API key — no --project flag needed anywhere. See Project Scoping when you reach that point.