Available Tools
The MCP server exposes 6 tools that cover the entire DataHippo platform. Each tool accepts a discriminator field that routes to the right operation, keeping the tool surface small while supporting 100+ distinct operations.
All tools operate within the context of the authenticated project.
Overview
| Tool | Discriminator | Purpose | Operations |
|---|---|---|---|
search | source | Find resources by text query | 3 |
get | resource | Retrieve a specific resource by type + ID | 21 |
list | resource | Browse/list resources with optional filters | 27 |
analyze | analysis | Metrics, analytics, comparisons, diagnostics | 18 |
execute | resource + action | Create, update, configure, deploy, test, run | 39 |
delete | resource | Remove a resource permanently | 5 |
search
Search across different data sources. Set source to select the search backend.
{ "source": "llm_requests", "query": "error handling", "model": "gpt-4o" }| Source | Description | Key Parameters |
|---|---|---|
llm_requests | Text search over LLM prompts and completions | query, limit?, model?, user_id?, session_id?, start_time?, end_time? |
logs | Search logs with cross-signal correlation | query?, level?, service?, trace_id?, time_range?, start_time?, end_time?, limit? |
web | Web search for real-time information | query, max_results? |
get
Retrieve a specific resource by type and ID. Set resource to select the resource type.
{ "resource": "trace", "trace_id": "abc123" }| Resource | Description |
|---|---|
trace | Distributed trace by ID |
session | LLM session with conversation metadata |
session_requests | All requests within an LLM session |
log | Single log entry by ID |
log_context | Surrounding log lines for context |
exception | Exception details with stack trace |
incident | Incident details and timeline |
incident_errors | Error events for a specific incident |
alert_rule | Alert rule configuration |
dashboard | Dashboard layout and widgets |
health_check | Health check configuration |
health_check_results | Recent health check probe results |
maintenance_window | Maintenance window details |
prompt_config | Prompt configuration |
prompt_version | Specific prompt version |
rollout | Rollout details |
rollout_metrics | Rollout performance metrics |
profile | Performance profile |
project | Project details |
request_scores | LLM request quality scores |
gateway_settings | LLM gateway settings |
list
Browse and list resources with optional filters. Set resource to select what to list.
{ "resource": "traces", "status": "error", "sort_by": "start_time", "limit": 10 }| Resource | Description |
|---|---|
traces | Distributed traces. Filters: status (error/ok), service, environment, service_version, http_method, http_route, search, start_time, end_time, sort_by (start_time/duration), sort_order (asc/desc), limit |
services | Monitored services |
service_versions | Service deployment versions |
sessions | LLM sessions |
exceptions | Exception groups |
incidents | Incidents |
api_endpoints | Auto-discovered API endpoints |
api_endpoint_errors | Errors for a specific API endpoint |
alert_rules | Alert rules |
alerts | Triggered alerts |
notification_channels | Notification channels |
dashboards | Dashboards |
dashboard_templates | Dashboard templates |
widgets | Widgets on a dashboard |
health_checks | Health checks |
maintenance_windows | Maintenance windows |
integrations | LLM provider integrations |
prompt_configs | Prompt configurations |
prompt_versions | Prompt versions for a config |
rollouts | Prompt rollouts |
profiles | Performance profiles |
service_profiles | Profiles for a specific service |
projects | Projects |
api_keys | API keys |
llm_scores | LLM quality scores |
llm_pricing | LLM model pricing (internal) |
metric_names | Available OpenTelemetry metric names. Filters: prefix?, limit?. Returns name, type, unit, label keys. Use before querying otel_metrics. |
analyze
Run analytics, queries, comparisons, and diagnostics. Set analysis to select the type.
{ "analysis": "llm_overview", "start_date": "2026-04-01", "end_date": "2026-04-07" }LLM Metrics
| Analysis | Description |
|---|---|
llm_overview | Gateway overview metrics (requests, latency, errors, costs) |
llm_model_metrics | Per-model LLM metrics breakdown |
llm_cost_daily | Daily LLM cost breakdown |
llm_user_metrics | Per-user LLM usage metrics |
Custom Data Queries
| Analysis | Description |
|---|---|
widget_query | Custom ClickHouse query on observability data |
otel_metrics | Query OpenTelemetry metrics by name. Params: metric_name (required), from?, to?, step?, time_aggregation?, space_aggregation?, filters?, group_by?. Use list metric_names first. |
LLM Testing
| Analysis | Description |
|---|---|
playground | Run a prompt in the LLM playground |
compare_models | Compare up to 5 LLM models side-by-side |
Service Diagnostics
| Analysis | Description |
|---|---|
compare_versions | Compare two service deployment versions |
compare_profiles | Compare performance profiles |
detect_faults | Detect faulty deployments via anomaly detection |
root_cause | AI-powered root cause analysis for an exception |
Project & Billing Analytics
| Analysis | Description |
|---|---|
project_stats | Project-level statistics |
endpoint_summary | API endpoint summary metrics |
usage | Overall platform usage |
usage_by_project | Usage broken down by project |
usage_forecast | Forecast future usage and costs |
budget_status | Budget status and spend tracking |
execute
Create, update, configure, deploy, test, and run operations. Provide resource, action, and params.
{
"resource": "alert_rule",
"action": "create",
"params": {
"name": "High Error Rate",
"query_config": { "metric_name": "http.server.errors", "..." : "..." },
"threshold": 5.0
}
}Resources and Actions
| Resource | Actions |
|---|---|
project | create, update, create_api_key |
prompt | create_config, update_config, create_version, deploy, promote, complete, pause, rollback |
gateway | update_settings |
integration | configure, update, test, create_secret_slot |
dashboard | create, create_from_template, update, create_widget, update_widget |
alert_rule | create, update |
health_check | create, update |
maintenance_window | create, update |
notification_channel | configure, update, test, configure_servicenow |
cloud_integration | configure_aws, configure_gcp, configure_azure, configure_oci |
auth_provider | configure |
github | list_installations, list_commits, get_commit, list_directory, read_file, search_code, link_repo, unlink_repo |
exception | update_status |
llm_score | submit |
GitHub
The github resource delegates to a multi-action handler. The action field selects the operation, and params contains action-specific fields:
| Action | Parameters | Description |
|---|---|---|
list_installations | — | List GitHub App installations available to the organization |
list_commits | branch?, limit? | List recent commits (default 10, max 30) |
get_commit | sha | Get details for a specific commit |
list_directory | path?, ref? | List files and subdirectories at a path |
read_file | path, ref? | Read the contents of a file |
search_code | query | Search for code by keyword or symbol |
link_repo | repository_url | Link the project to a GitHub repository |
unlink_repo | — | Unlink the project from its repository |
delete
Remove a resource permanently. Set resource to select what to delete. Each resource type has its own ID field.
{ "resource": "alert_rule", "rule_id": "abc123", "confirm": true }| Resource | Description |
|---|---|
integration | Delete an LLM provider integration |
alert_rule | Delete an alert rule |
health_check | Delete a health check |
maintenance_window | Delete a maintenance window |
widget | Delete a dashboard widget |