AgentKit
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| POSTGRES_URL | Yes | PostgreSQL connection URL for the KPI database | |
| ANTHROPIC_API_KEY | Yes | Anthropic API key for advanced analysis features |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| logging | {} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| query_kpisC | Return KPI metrics for a domain and period window. |
| get_company_healthB | Return composite company health index for a domain (or all). |
| detect_kpi_anomaliesC | Find anomalies in a domain's KPI history. |
| forecast_metricA | Forecast |
| list_available_metricsA | Discovery tool: list metrics, categories, and periods (metrics scoped to domain if given). |
| get_executive_summaryA | Synthesize health, KPIs, and anomalies into a one-shot executive summary. |
| list_annotationsA | Read annotations, most recent first. Optionally filter to one metric. Retracted annotations are excluded unless include_retracted is true. |
| annotate_metricA | Attach a durable note to a metric/period — e.g. an agent recording the root cause it identified for an anomaly, so the next reader sees the explanation inline. Effect: WRITE — mutates data. Pass dry_run=true to preview without committing. |
| retract_annotationA | Soft-delete an annotation (sets retracted_at). Irreversible from the agent's side and externally visible, so it is classed destructive and requires human approval. Effect: DESTRUCTIVE — mutates data. Pass dry_run=true to preview without committing. Requires a human-supplied approval_token. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| monthly_executive_briefing | Reusable prompt template for a monthly cross-domain executive briefing. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| kpi_finance_latest | Latest Finance KPI snapshot. |
| kpi_people_latest | Latest People KPI snapshot. |
| kpi_operations_latest | Latest Operations KPI snapshot. |
| kpi_customer_latest | Latest Customer KPI snapshot. |
| kpi_engineering_latest | Latest Engineering KPI snapshot. |
| kpi_growth_latest | Latest Growth KPI snapshot. |
| kpi_logistics_latest | Latest Logistics KPI snapshot. |
| kpi_esg_latest | Latest ESG KPI snapshot. |
| kpi_it_latest | Latest IT KPI snapshot. |
| kpi_security_latest | Latest Security KPI snapshot. |
TDQS
Scored across 9 tools
Each tool targets a distinct resource and action: raw KPI queries, health indices, anomaly detection, forecasting, metric discovery, executive summaries, and annotation CRUD. There is no meaningful overlap between query_kpis, get_company_health, and get_executive_summary because they return different levels of aggregation.
All tool names follow a clear verb_noun snake_case pattern: query, get, detect, forecast, list, annotate, retract. The naming style is uniform and predictable, with only trivial variation between query, get, and list as read verbs.
Nine tools is a well-scoped size for a KPI analytics and annotation server. Each tool has a clear purpose, and there is no redundancy or bloat.
The tool surface covers metric discovery, retrieval, anomaly detection, forecasting, health summaries, and the full annotation lifecycle of create, read, and soft-delete. There are no obvious dead ends; even the destructive retract action has a preview and approval path.