Brawler MCP Server (Python)
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
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": true
} |
| resources | {
"subscribe": false,
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| policiesB | Company policy injected into the agent system prompt. |
| open_ticketC | Ticket reminder injected as a user message. |
| localeC | Locale hint that takes a language code. |
| boomA | Context tool that fails so the client can observe a swallowed error. |
| echoA | Echo the given text back to the model. |
| addB | Add two integers and return structured output. |
| failC | Completion tool that raises so run_tool_call can observe a remote error. |
| groundedD | Reject empty answers. |
| score_lengthB | Score a non-empty answer by character length. |
| auditC | No-op post tool used to verify post-task listing and calls. |
| recordD | Post tool with multiple input fields. |
| pingC | Generic MCP ping; Brawler agents skip none tools. |
| legacy_searchD | Untagged FastMCP tool; Brawler treats it as completion. |
| styleC | Skill-tagged tool; Brawler treats unsupported tasks as none. |
| routeD | Router-tagged tool; Brawler treats unsupported tasks as none. |
| trainC | Training-tagged tool; Brawler treats unsupported tasks as none. |
| unknown_taskD |
|
| tagged_evalC | Evaluation via FastMCP tags only (no |
| broken_evalC | Evaluation tool that returns non-object text. |
| valid_no_scoreC | Evaluation tool that omits |
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 20 tools
Several tools overlap heavily: boom and fail both simulate errors, tagged_eval/broken_eval/valid_no_score are evaluation variants, and style/route/train are all tagged tools treated as unsupported tasks. Descriptions help somewhat, but an agent would struggle to choose between near-duplicate test fixtures.
Names mix nouns (policies, locale, audit, record, ping) with bare verbs (boom, echo, add, fail) and inconsistent snake_case phrases (legacy_search, unknown_task, valid_no_score). There is no predictable verb_noun pattern, making the surface feel ad hoc.
At 20 tools the server is on the heavy side, especially since many are narrowly scoped test fixtures rather than distinct capabilities. The count is not extreme, but several tools could be consolidated without losing coverage.
The apparent domain is MCP behavior testing, and while it covers errors, tagging, post tools, and evaluation variants, the surface is a grab bag of edge cases rather than a complete lifecycle. Obvious gaps exist around real resource operations, and the set does not form a coherent API.