Loading ToolBench
AI BVF: score AI portfolios Stop/Fix/Accelerate with decision confidence and pace-layer drag.
aibvf-mcp demonstrates solid definition quality with 13 well-named, domain-specific tools covering AI portfolio assessment. All tools have descriptions (avg 120 chars) and explicit input schemas with typed parameters. Naming follows verb_noun convention (assess_, score_, sequence_, diagnose_). However, output schemas are not documented in the visible code, parameter descriptions lack actionable constraints (enums, ranges, formats), and error handling guidance is absent. No tool annotations (readOnlyHint, destructiveHint) despite all tools being READ_ONLY. Schemas are present but sparse, many parameters lack detailed descriptions of valid ranges or formats.
Assemble a portfolio from a list of initiatives and return the portfolio structure with metadata.
Assess an AI initiative: extract revenue, validate inputs, infer readiness, and return structured assessment with confidence.
Calculate pace-layer drag for an initiative and return the drag decomposition.
Diagnose a business process and return intervention recommendations with baseline cost and net saving estimates.
Get benchmark data for an industry, function, and AI tier combination.
Infer the readiness level of an organization based on provided signals.
Output schemas not documented. Tools return structured results but LLMs cannot see what fields to expect, forcing them to guess downstream field names and risk extraction errors.
Parameter descriptions lack actionable constraints. 'Industry classification' and 'AI tier classification' do not specify valid enum values, ranges, or formats. LLMs cannot self-correct invalid input.
No error handling guidance. Tools do not document what errors can occur, whether they are retryable, or what the LLM should do next. A validation failure returns nothing actionable.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 63 | 2026-07-28+ | v2 |
List all valid values for industry, function, AI tier, and readiness enums.
Map user-provided strings to canonical AI BVF taxonomy values (industry, function, tier, readiness).
Recommend improvements to move a Fix initiative toward Accelerate.
Score a single AI initiative and return Stop/Fix/Accelerate verdict with decision confidence and net value range.
Score a portfolio of AI initiatives and return verdicts, confidence, and net value for each.
Sequence a portfolio of initiatives by priority and return the sequenced list with rationale.
Validate a portfolio manifest against the AI BVF schema and return validation errors if any.
Tool annotations missing. All 13 tools are READ_ONLY but lack readOnlyHint annotation in schema. This prevents clients from optimizing caching and retry logic.
Complex nested objects (scores, work_architecture, signals) lack field-level documentation. LLMs cannot determine which nested fields are required, their types, or valid values.