jev-cognitive-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| JEV_BRIDGE_DIR | No | Path to the local JevBridge runtime directory. Defaults to %LOCALAPPDATA%\JevBridge\runtime on Windows or ~/.local/share/JevBridge/runtime on Linux/macOS. Required only for tools that invoke the external JevBridge runtime; offline deterministic tools do not require it. |
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 | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| jev_guardA | Provide advisory risk screening for a shell command using Layer-0 deterministic regex rules and (unless reflex_only=true) System-1 probabilistic evaluation. Does not intercept host execution. |
| jev_routeB | Provide advisory task complexity and model tier recommendation (requires configured JevBridge runtime). |
| jev_verifyB | Review caller-supplied task specification, change summary, and test logs for potential evidence gaps (does not run tests or read git diffs directly; requires JevBridge runtime). |
| jev_rerankA | Rank up to 96 candidate text items against a query in batches of 12 (requires JevBridge runtime). |
| jev_arbitrateB | Run the offline deterministic 5-factor Meta-Controller to recommend Layer-0 Reflex, System-1 Intuition, System-2 Deep Reasoning, or Human Gate. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 5 tools
Each tool has a distinct core action: risk screening (guard), complexity/tier routing (route), text ranking (rerank), evidence-gap review (verify), and layer arbitration (arbitrate). The only mild overlap is between route (model tier) and arbitrate (reasoning layer), which could be confused, but descriptions distinguish them adequately.
All tools follow a strict 'jev_' prefix plus a single verb pattern (guard, route, rerank, verify, arbitrate). Fully predictable and consistent throughout.
Five tools is well-scoped for a cognitive decision-support server, with each tool earning its place as a distinct cognitive operation. No redundancy or filler.
The surface covers the main cognitive advisory steps: screen, route, rank, verify, and arbitrate. Minor gaps exist (e.g., no explicit memory/state or feedback tool), but core workflows for a decision-support layer are covered.