Loading ToolBench

Reading the current corpus…

tacit MCP server · C · ToolBench

tacit

A Model Context Protocol server for safe Scala code execution with capability-scoped API access, file system operations, process execution, and network requests.

2026-07-28+STDIO6 tools

63/100

Definition Quality72

Protocol Readiness50

Supportability55

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence · evidence: stateless requests

Current-spec patterns detected

Summary

TACIT MCP defines 6 tools with complete JSON schemas and descriptions. Naming follows verb_noun convention (execute_scala, create_repl_session, delete_repl_session, execute_in_session, list_sessions, show_interface). All tools have descriptions (avg 115 chars, within 10-1024 range). However, parameter descriptions are minimal or absent: 'code' param lacks detail on what Scala code is valid; 'session_id' descriptions are terse. No output schemas documented, LLMs cannot predict return structure. Error handling is not visible in tool definitions. The show_interface tool is a discovery pattern but lacks guidance on when to call it. Session management tools (create/delete/list/execute_in_session) form a coherent subsystem but lack explicit state management warnings.

Tools (6)

create_repl_sessionread onlysource verified73/100

Create a new Scala REPL session. Returns a session ID that can be used for subsequent executions.

delete_repl_sessionreversiblesource verified73/100

Delete a Scala REPL session by its ID.

execute_in_sessionwritesource verified73/100

Execute Scala code in an existing REPL session. The session maintains state between executions. The library API is pre-loaded, and all functions defined in `Interface` are available.

execute_scalawritesource verified73/100

Execute a Scala code snippet and return the output. This is stateless, each execution is independent. The library API is pre-loaded, and all functions defined in `Interface` are available.

list_sessionsread onlysource verified72/100

List all active REPL session IDs.

show_interfaceread onlysource verified65/100

Show the full capability-scoped API available in the REPL. Call this first to understand what methods you can use. You must only use the provided interface to interact with the system.

Issues (5)

HIGHtool ↗

No output schemas documented. LLMs cannot predict return structure (e.g., does execute_scala return {result: string} or {output: string, error?: string}?). This forces agents to guess and risks parsing failures.

execute_scalacreate_repl_sessionexecute_in_sessionlist_sessionsshow_interface

HIGHtool-description ↗

Parameter descriptions are minimal. 'code' parameter lacks guidance on valid Scala syntax, library availability, or constraints. 'session_id' descriptions are terse. LLMs need explicit format/constraint hints (e.g., 'UUID format', 'max 1000 lines', 'no file I/O').

execute_scaladelete_repl_sessionexecute_in_session

HIGHrecovery-guide ↗

No error handling guidance in tool definitions. If execute_scala fails (syntax error, timeout, security violation), LLMs have no recovery path. Error responses should guide: 'Syntax error at line 5. Check Scala syntax.' or 'Execution timeout (>30s). Simplify code or use session for incremental work.'

execute_scala

Recommendations

Document output schemas for all tools. Example: execute_scala returns {success: boolean, output?: string, error?: string, executionTimeMs: number}. Include this in tool descriptions or a separate schema reference.
Expand parameter descriptions with format/constraint hints. For 'code': 'Scala code snippet (max 10,000 chars). Pre-loaded: Interface.scala methods. No file I/O or network access. Syntax errors return detailed error messages.' For 'session_id': 'UUID of an active REPL session (created via create_repl_session). Invalid IDs return 404 with available session list.'
Add error handling guidance to execute_scala and execute_in_session descriptions. Example: 'On syntax error, returns error message with line number. On timeout (>30s), returns timeout error, use session mode for long-running code. On security violation (file I/O attempt), returns permission denied.'
Declare idempotency via toolAnnotations (idempotentHint: true for create_repl_session if it returns same ID on retry) or explicit description: 'Idempotent: calling twice with same parameters returns the same session_id.'
Enhance show_interface description: 'Call this first to discover available API methods in the REPL. Returns the full Interface.scala source. Methods are pre-loaded in all code executions. Use this to understand what system interactions are available before writing code.'

Score history

Overall score trend

First recorded score · v2 rubric

63/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	C	63	2026-07-28+	v2