agent-context-router
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ACR_LOG | No | Path to the load log. Defaults to ~/.local/state/agent-context-router/loads.jsonl. | |
| ACR_REGISTRY | No | Path to the registry YAML file. If unset, defaults to ~/.config/agent-context-router/modules.yaml. | |
| ACR_CACHE_DIR | No | Directory for sync state. Defaults to ~/.cache/agent-context-router. | |
| AGENT_MACHINE | No | Machine profile, either work or personal. Defaults to work. | work |
| ACR_SYNC_WINDOW | No | Seconds between fetches of the same repo. | 1800 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
Loads agent policy modules on demand.
The always-loaded AGENTS.md lists each module with a trigger. Before doing the work a trigger names, call load_module with that id. Use list_modules to see every module, list_modules with a dir for that directory's index card, and search_context when no trigger obviously matches. Each load reports the repo commit it read and any reason the local copy couldn't be updated.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| logging | {} |
| prompts | {
"listChanged": true
} |
| resources | {
"subscribe": false,
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| list_modulesA | List every policy module available on this machine, with the trigger that says when to load it. |
| load_moduleB | Load a policy module's full text. Call before the work its trigger names. |
| search_contextB | Find modules relevant to a task when no trigger obviously matches. |
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 3 tools
The three tools cover distinct retrieval modes: full enumeration (list_modules), keyed fetch (load_module), and fuzzy discovery (search_context). The descriptions explicitly disambiguate the two discovery paths — list is for trigger-based lookup, search is the fallback when no trigger matches — so an agent can pick correctly.
All three names follow a consistent verb_noun snake_case pattern (list_modules, load_module, search_context). The verbs are semantically meaningful and map cleanly onto the action each tool performs.
Three tools is on the thin side but each earns its place in a discover-load-search workflow. Nothing redundant is present, though the surface is small enough that a single missing operation would be noticeable.
The surface covers the full read lifecycle: enumerate modules, fuzzy-find one, and load its text. Minor gaps exist — no way to unload/release a loaded module or check version/freshness of the policy corpus — but these are workarounds an agent can live without.