MCP server for BASE structured knowledge, routing, and mediated actions. Exposes BASE agents to any AI platform via the Model Context Protocol. Router primitives, one lightweight compatibility bootstrap (load_agent), two transports (stdio, HTTP), zero business orchestration.
18 tools with significant quality gaps. Most tools have basic descriptions (50-150 chars) but lack depth on WHEN to use them vs similar tools. Input schemas present but sparse: many parameters lack descriptions or type constraints. No output schemas documented. Error handling absent, tools return success/failure with no recovery guidance. Tool names follow verb_noun pattern (good), but composition issues exist: invoke_tool, propose_change, commit_change suggest multi-step workflows that should be atomic. No tool annotations (readOnlyHint/destructiveHint). Security concerns: propose_change and commit_change accept generic 'object' params with no validation hints. Descriptions do not explain prerequisites, dependencies, or when to call discovery tools first.
Access a BASE resource with permission checking
Commit a proposed change to a BASE resource
Discover and inventory BASE resources in the project
Get the context and capabilities for a specific agent
Get the routing map for BASE agents and processes
Get warnings and issues for the BASE workspace
Invoke a tool within a BASE agent context
List all available BASE agents in the project
invoke_tool accepts generic 'input' object with no schema or validation. LLM cannot infer what parameters a tool accepts without calling get_agent_context first. Breaks tool composition pattern.
propose_change and commit_change accept 'change' and 'proposal_id' as opaque objects/strings with no schema. No validation hints, no error recovery guidance. Agents cannot reason about what changes are valid.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., agent_id, resource_id). Breaks tool chaining.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | <=2025-11-25 | v2 |
List all markers (tags/labels) in the BASE project
Load a BASE agent by name from the project root
Open and read the content of a BASE resource
Promote a resource to higher visibility or priority
Propose a change to a BASE resource
Report friction or issues encountered during agent execution
Resolve the BASE configuration for the project
Route a request to the appropriate BASE agent or process
Search BASE resources by query
Get the walk policy for resource discovery
Descriptions lack WHEN/WHY context. E.g., 'Get the routing map' does not explain when to call it vs route_request, or what structure it returns. LLMs cannot disambiguate similar tools.
No error handling or recovery guidance. Tools return success/failure with no actionable next steps. LLMs cannot self-correct or retry intelligently.