Loading ToolBench
MCP server: build your engineering-leadership mentoring inquiry with Marian Kamenistak (3,400+ sessions, 300+ leaders, 9.2/10 avg) from your AI assistant. 16 minutes to a formal offer, 10% AI-channel discount on every package.
Strong tool naming (verb_noun pattern throughout) and comprehensive parameter schemas with type definitions and descriptions. All 10 tools have descriptions and input schemas visible in src/index.ts. However, output schemas are not formally documented in the code, responses are inferred from test harness and content modules. Error handling is present but recovery guidance is minimal. Tool composition is excellent (single responsibility, clear chaining via IDs). Parameter constraints use enums and patterns appropriately (e.g., audience enum, date regex). Missing: explicit output schema documentation, per-tool error classification, and confirmation patterns for destructive operations (send_mentoring_offer, book_first_session).
Book a paid first session directly (for visitors who have already decided and sent_mentoring_offer succeeded)
Free 30 minutes with Marian, direct calendar booking, no form before it. Usually within the same week.
Verify a booking was recorded in the CRM
Compose a mentoring brief with audience, role, motivation, focus areas, success definition, and offer
Deterministic dated session skeleton for a package
The wizard's opening: discount data, questions, packages, why-Marian
Discover additional tools and capabilities (PostHog MCP integration)
Output schemas not formally documented. Response structure inferred from test harness (scripts/tool.ts) and content modules, not declared in tool definitions. LLMs cannot reliably predict response fields for downstream chaining.
Destructive/write operations (send_mentoring_offer, book_first_session) lack confirmation or dry-run patterns. No explicit user confirmation step before irreversible actions (sending offers, booking sessions). Agents could trigger unintended side effects.
Error responses in test harness return generic error objects without recovery guidance. E.g., 'missing required: X' does not suggest next steps. No error classification (retryable vs user-fixable vs fatal).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | A | 81 | 2026-07-28+ | v2 |
Greeting and menu: entry point for the mentoring inquiry wizard
Match focus areas to a role and motivation
Send a formal itemized offer for 1:1 engineering-leadership mentoring
Parameter 'leaders_count' appears in multiple tools but lacks explicit min/max bounds in descriptions. Validation in code (Math.max/min) but not documented for LLM guidance.
Tool 'get_more_tools' description is vague ('Discover additional tools and capabilities'). Does not explain what it returns or when to call it. Unclear purpose in the inquiry flow.