A collection of quantitative analysis and optimization algorithm skills for e-commerce, manufacturing, pricing, ranking, risk assessment, and supply chain management
13 tools with complete input schemas and descriptions. Naming follows verb_noun pattern (compute, update, process). Descriptions are substantive (100-250 chars avg), explaining what each tool does and when to use it. However, output schemas are not documented in the MCP tool definitions, only visible in the Python source code comments. Parameter descriptions are present but lack explicit constraints (min/max, enums where applicable). No error handling guidance in tool descriptions. No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All tools are READ_ONLY by risk classification, but this is not declared in the MCP schema.
Compute Altman Z-Score for bankruptcy prediction. Calculates Z-score and assigns zone (safe/grey/distress). Supports original, private firm, and non-manufacturing variants.
Compute arc (midpoint) elasticity of demand between two price-quantity points. Calculates price elasticity, classification (elastic/inelastic/unit elastic), and revenue impact.
Rank items by Bayesian average (IMDB-style weighted rating). Pulls items with few ratings toward the global mean to prevent small-sample extremes from dominating rankings.
Score documents against query using BM25 (Best Matching 25) ranking algorithm. Computes IDF-weighted term frequency scores with document length normalization.
Compute Process Capability Index (Cpk) and related metrics. Calculates Cp, Cpk, Cpu, Cpl, PPM defective, and process capability assessment based on specification limits and process statistics.
Process a sequence of matches and update Elo ratings in order. Returns final ratings and leaderboard.
Output schemas not documented in MCP tool definitions. Python source includes detailed return structures (e.g., bm25.py returns {query, parameters, n_docs, results}), but these are not exposed in the MCP schema. LLMs cannot plan downstream operations without knowing what fields to expect.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All 13 tools are marked Risk: READ_ONLY in metadata, but this is not declared via MCP tool annotations. Agents cannot infer safety properties from the schema alone.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
Update Elo ratings after a match. Computes expected scores and rating deltas for both players. Maintains zero-sum property where winner gains equal loser losses.
Compute Economic Order Quantity (EOQ) and related inventory metrics. Calculates optimal order quantity, ordering frequency, holding costs, and reorder point.
Compute optimal order quantity for single-period inventory problem using newsvendor model. Calculates critical ratio, optimal quantity, and expected profit.
Compute point elasticity of demand at a specific price-quantity point given the marginal dQ/dP. Calculates elasticity and classification.
Compute safety stock and reorder point with combined demand and lead-time variability. Accounts for both demand and lead-time uncertainty.
Compute Wilson score confidence interval for a single binomial proportion. Returns lower and upper bounds that penalize items with few observations.
Rank a list of items by Wilson score lower bound. Prevents the '5/5 stars with 1 review' problem by penalizing items with few observations.
Parameter constraints not fully specified. Numeric parameters (k1, b, sd, price, cost, etc.) lack explicit min/max bounds in descriptions. E.g., cpk_compute.sd says 'must be positive' but no min value stated; eoq_compute.annual_demand says 'must be positive' but no upper bound. LLMs may pass absurd values.
No error recovery guidance in tool descriptions. Python source includes validation (e.g., bm25.py raises ValueError for empty documents, invalid k1/b), but MCP tool descriptions do not explain what errors are possible or how to recover. Agents cannot self-correct on failure.
Enum constraints not declared in schema. altman_z_compute.variant accepts 'original', 'private', 'non_manufacturing' but is typed as string, not enum. wilson_score_compute_interval.confidence accepts specific values (0.80, 0.90, 0.95, 0.99, 0.999) but not declared as enum. LLMs may hallucinate invalid values.