Redirecting to documentation...
If you are not redirected automatically, click here.
Model Context Protocol server for accessing Respan data from your AI assistant
Use the hosted endpoint where each user provides their own API key:
{ "mcpServers": { "respan": { "url": "https://mcp.respan.ai/api/mcp", "headers": { "Authorization": "Bearer YOUR_RESPAN_API_KEY" } } } }
Run locally for development - simplest setup:
git clone https://github.com/respanai/respan-mcp.git cd respan-mcp npm install && npm run build
Configure your MCP client (~/.cursor/mcp.json):
~/.cursor/mcp.json
{ "mcpServers": { "respan": { "command": "node", "args": ["/path/to/respan-mcp/dist/lib/index.js"], "env": { "RESPAN_API_KEY": "your_api_key" } } } }
Deploy to Vercel with API key stored as environment variable. No client-side key needed:
git clone https://github.com/respanai/respan-mcp
vercel deploy --prod
RESPAN_API_KEY
Configure your MCP client:
{ "mcpServers": { "respan": { "url": "https://your-project.vercel.app/mcp" } } }
Everyday tools are listed at connect; tool_search finds the rest by resource and action words and tool_run runs them.
tool_search
tool_run
log_list
log_get
log_summary
trace_list
trace_get
metric_summary
metric_timeseries
error_group_list
behavior_summary
prompt_list
All list tools support powerful filtering:
{ "cost": {"operator": "gt", "value": [0.01]}, "model": {"operator": "", "value": ["gpt-4"]}, "customer_identifier": {"operator": "contains", "value": ["user"]}, "metadata__session_id": {"operator": "", "value": ["abc123"]} }
Operators: "" (equal), not, lt, lte, gt, gte, contains, icontains, startswith, endswith, in, isnull
""
not
lt
lte
gt
gte
contains
icontains
startswith
endswith
in
isnull