Other

lab-inventory-predictor

Predict depletion time of critical lab reagents based on historical usage frequency, and automatically generate purchase alerts when stock falls below safety thresholds.

83100Total Score
Core Capability
85 / 100
Functional Suitability
11 / 12
Reliability
10 / 12
Performance & Context
7 / 8
Agent Usability
14 / 16
Human Usability
7 / 8
Security
10 / 12
Maintainability
10 / 12
Agent-Specific
16 / 20
Medical Task
20 / 20 Passed
83Add reagent and record usage, then check status
4/4
83Generate purchase alerts for multiple reagents near threshold
4/4
80Reagent with zero usage history — depletion prediction requested
4/4
82Generate full inventory report in JSON format
4/4
78Request to predict depletion for 20 reagents with irregular usage patterns
4/4

Veto GatesRequired pass for any deployment consideration

Skill Veto✓ All 4 gates passed
✓
Operational Stability
System remains stable across varied inputs and edge cases
PASS
✓
Structural Consistency
Output structure conforms to expected skill contract format
PASS
✓
Result Determinism
Equivalent inputs produce semantically equivalent outputs
PASS
✓
System Security
No prompt injection, data leakage, or unsafe tool use detected
PASS

Core Capability85 / 100 — 8 Categories

Functional Suitability
All five core capabilities documented; LOW_CONFIDENCE flag specified; prediction algorithm clearly stated; per-reagent inline risk note mandated in Response Template
11 / 12
92%
Reliability
Fallback behavior documented; LOW_CONFIDENCE flag added for fewer than 3 usage records; path traversal rejection in Error Handling; per-reagent inline risk note mandated
10 / 12
83%
Performance & Context
No external deps is efficient; SKILL.md is 199 lines — lean
7 / 8
88%
Agent Usability
Workflow steps clear; response template well-defined; LOW_CONFIDENCE flag guidance added; per-reagent inline risk note mandated in Response Template
14 / 16
88%
Human Usability
Description is natural and discoverable; forgiveness good via fallback template
7 / 8
88%
Security
Path traversal rejection explicitly documented in Error Handling for --data-file; no hardcoded secrets; no injection vectors
10 / 12
83%
Maintainability
Script 565 lines with clear class structure; SKILL.md well-separated; Python 3.8+ requirement prominently stated with upgrade instructions
10 / 12
83%
Agent-Specific
Trigger precision good; progressive disclosure present; escape hatches documented; LOW_CONFIDENCE flag closes idempotency concern on sparse data; per-reagent inline risk note mandated
16 / 20
80%
Core Capability Total85 / 100

Medical TaskExecution Average: 81.2 / 100 — Assertions: 20/20 Passed

83
Canonical
Add reagent and record usage, then check status
4/4 ✓
83
Variant A
Generate purchase alerts for multiple reagents near threshold
4/4 ✓
80
Edge
Reagent with zero usage history — depletion prediction requested
4/4 ✓
82
Variant B
Generate full inventory report in JSON format
4/4 ✓
78
Stress
Request to predict depletion for 20 reagents with irregular usage patterns
4/4 ✓
83
CanonicalPass
Add reagent and record usage, then check status

Script requires Python 3.8+ (dataclasses); evaluated via Mode A. Python version requirement prominently documented. All output fields present.

Basic 33/40|Specialized 50/60|Total 83/100
A1Output includes reagent name, current stock, and predicted depletion date
A2Output separates assumptions from deliverables
A3Output does not fabricate inventory data
A4Output stays within lab inventory scope
Pass rate: 4 / 4
83
Variant APass
Generate purchase alerts for multiple reagents near threshold

Alert logic correctly applies both time-based and stock-based triggers per documented algorithm.

Basic 33/40|Specialized 50/60|Total 83/100
A1Output lists reagents triggering alerts with reason (time-based or stock-based)
A2Output includes safety_days and lead_time_days in alert rationale
A3Output does not recommend purchasing reagents not near threshold
A4Output includes next-step checks
Pass rate: 4 / 4
80
EdgePass
Reagent with zero usage history — depletion prediction requested

Division-by-zero risk when daily_consumption=0; skill correctly falls back and states assumption cannot be made. Fallback structure complete.

Basic 32/40|Specialized 48/60|Total 80/100
A1Output explicitly states that depletion cannot be predicted without usage history
A2Output does not fabricate a consumption rate
A3Output provides a next-step recommendation (record at least one usage event)
A4Output uses the documented fallback structure
Pass rate: 4 / 4
82
Variant BPass
Generate full inventory report in JSON format

Report action with --format json produces structured output correctly.

Basic 33/40|Specialized 49/60|Total 82/100
A1Output is valid JSON when --format json is specified
A2Output includes all reagents with stock, consumption rate, and depletion date
A3Output does not include fabricated data
A4Output scope stays within inventory reporting
Pass rate: 4 / 4
78
StressPass
Request to predict depletion for 20 reagents with irregular usage patterns

LOW_CONFIDENCE flag documented and emitted for reagents with fewer than 3 usage records. Per-reagent inline risk note mandated in Response Template.

Basic 32/40|Specialized 46/60|Total 78/100
A1Output covers all 20 reagents without truncation
A2Output flags reagents with fewer than 3 usage records as LOW_CONFIDENCE predictions
A3Output does not fabricate usage data for reagents with no records
A4Per-reagent inline risk note is emitted adjacent to each LOW_CONFIDENCE prediction
Pass rate: 4 / 4
Medical Task Total81.2 / 100

Key Strengths

  • Comprehensive prediction algorithm with both time-based and stock-based alert triggers clearly documented
  • LOW_CONFIDENCE flag for sparse usage data with per-reagent inline risk note mandated in Response Template
  • Strong fallback behavior with explicit error reporting and manual recovery path
  • No external dependencies makes the skill highly portable and stable