UNITARES
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| REDIS_URL | No | Redis connection URL (optional, e.g., redis://redis:6379/0) | |
| DB_BACKEND | Yes | Database backend (e.g., postgres) | |
| DB_AGE_GRAPH | Yes | Apache AGE graph name (e.g., governance_graph) | |
| DB_POSTGRES_URL | Yes | PostgreSQL connection URL (e.g., postgresql://postgres:postgres@localhost:5432/governance) | |
| UNITARES_DISABLE_ODE | No | Set to '1' to disable the ODE math model | 0 |
| UNITARES_KNOWLEDGE_BACKEND | Yes | Knowledge backend (e.g., age) |
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": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| start_sessionA | Primary workflow name for starting a session; implemented by onboard(). Common case: force_new=true alone — save the returned uuid and client_session_id. parent_agent_id is for a real handoff from an exited predecessor, not for a session sharing the workspace with one still running. |
| sync_stateC | Primary workflow name for checking in state; implemented by process_agent_update(). EISV fields: E=Energy [0,1] (mixed-provenance capacity estimate); I=Information Integrity [0,1] (mixed-provenance calibration estimate); S=Entropy [0,1] (drift from the agent's own normal); V=Valence [-1,1] (EMA-smoothed E-I imbalance; positive=motion outruns integrity, negative=integrity outruns motion). |
| check_working_stateB | Primary workflow name for reading current EISV state; implemented by get_governance_metrics(). EISV fields: E=Energy [0,1] (mixed-provenance capacity estimate); I=Information Integrity [0,1] (mixed-provenance calibration estimate); S=Entropy [0,1] (drift from the agent's own normal); V=Valence [-1,1] (EMA-smoothed E-I imbalance; positive=motion outruns integrity, negative=integrity outruns motion). |
| search_shared_memoryC | Primary workflow name for memory search; implemented by knowledge(action='search'). |
| store_findingA | Primary workflow name for recording a finding to shared memory; implemented by knowledge(action='store'). Use for durable knowledge — a discovery, root cause, or correction. For task/tool/test outcomes use record_result instead. |
| update_findingA | Primary workflow name for revising a finding already in shared memory; implemented by knowledge(action='update'). Needs the discovery_id; set status when the finding is resolved or superseded. |
| record_resultB | Primary workflow name for recording outcomes; implemented by outcome_event(). EISV fields: E=Energy [0,1] (mixed-provenance capacity estimate); I=Information Integrity [0,1] (mixed-provenance calibration estimate); S=Entropy [0,1] (drift from the agent's own normal); V=Valence [-1,1] (EMA-smoothed E-I imbalance; positive=motion outruns integrity, negative=integrity outruns motion). |
| request_reviewB | Primary workflow name for structured review; implemented by dialectic(action='request'). The issue description is reused as the thesis by default, so a reviewer answers or a verdict returns without duplicating the brief. Pass use_brief_as_thesis=false for the explicit two-call flow. |
| health_checkA | Quick health check - returns system status, version, component health, and continuity mode. Useful for monitoring and operational visibility. |
| describe_toolB | 📖 Get full details for a specific tool. Deep dive into any tool. |
| consultA | Primary advisory model-help surface. Choose purpose, standard or thorough effort, and whether external processing is allowed; routing and any explicit local degradation are reported. It returns advisory evidence, not governed judgment. Use response_mode='full' for route diagnostics. consult never creates a review verdict; use request_review for that. |
| identityC | Inspect or re-bind your identity. |
| knowledgeC | Unified knowledge graph operations: store, search, get, list, update, details, note, cleanup, synthesize, stats, supersede, audit. |
| self_recoveryB | Unified self-recovery for stuck/paused agents. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| UNITARES Governance SKILL | Governance framework orientation document for agents |
This server cannot be deployed
TDQS
Scored across 14 tools
Most workflow tools are distinct and well-described, such as request_review versus consult and record_result versus store_finding. However, the generic knowledge tool duplicates the search/store/update surfaces of search_shared_memory, store_finding, and update_finding without clear guidance on when to prefer the specialized wrappers, creating real selection ambiguity.
Most tools follow a readable verb_noun snake_case pattern like record_result, start_session, and store_finding. The pattern is broken by knowledge, identity, consult, and self_recovery, which are bare nouns/verbs or compound nouns, making the naming style mixed but still understandable.
14 tools is a reasonable size for a broad agent lifecycle server, and most tools represent distinct concerns. The count is slightly heavier than necessary because the knowledge tool overlaps with three specialized wrappers, but overall it is still well-scoped.
The set covers session start, state read/update, result recording, finding storage/update/search, structured review, consultation, health checks, identity, and self-recovery—a fairly complete operational lifecycle. Minor gaps include no explicit session end/close tool and no direct result retrieval, but these are likely workaroundable.