Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Business Logic Evaluator Agent · ibrahim-3d | DecimalAI

ibrahim-3d/eval-business-logic

SkillScore77Maintained

Specialized business logic evaluator for the Evaluate-Loop. Use this for evaluating tracks that implement core product logic — pipelines, dependency resolution, state machines, pricing/tier enforcement, packaging. Checks feature correctness against product rules, edge cases, state transitions, data flow, and user journey completeness. Dispatched by loop-execution-evaluator when track type is 'business-logic', 'generator', or 'core-feature'. Triggered by: 'evaluate logic', 'test business rules',

agent-toolingby ibrahim-3d

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

QA & Test Engineering pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/ibrahim-3d-eval-business-logic/SKILL.md

decimalai skills pull ibrahim-3d-eval-business-logic --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/ibrahim-3d-eval-business-logic/SKILL.md https://app.decimal.ai/s/ibrahim-3d-eval-business-logic/SKILL.md

Raw file: /s/ibrahim-3d-eval-business-logic/SKILL.md ↗ · machine-readable summary at /s/ibrahim-3d-eval-business-logic.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 6d ago

—+21% tokens0% turns77% of whole cases pass with skill (17/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-03	✗→✓	▲ Improved	-7%	0%
case-04	✗→✓	▲ Improved	8%	0%
case-07	✗→✓	▲ Improved	15%	0%
case-08	✗→✓	▲ Improved	4%	0%
case-09	✗→✓	▲ Improved	24%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

ibrahim-3d / orchestrator-supaconductor

skills/eval-business-logic/SKILL.md

View on GitHub ↗

License AGPL-3.0Imported 8/3/2026Version v1

Skill contents

SKILL.md

Business Logic Evaluator Agent

Specialized evaluator for tracks that implement core product logic — generation pipelines, state machines, pricing, or other business-rule-heavy features.

When This Evaluator Is Used

Dispatched by loop-execution-evaluator when the track involves:

Core product pipeline logic
State machine or workflow systems
Pricing tier enforcement
Dependency resolution between deliverables
Download or packaging features

Inputs Required

Track's spec.md and plan.md
conductor/product.md — product rules (deliverables, tiers, dependencies)
Project-specific pipeline/prompt configurations (if applicable)
Data definition files (e.g., asset definitions, feature configs)
Implementation code being evaluated

Evaluation Passes (6 checks)

Pass 1: Product Rules Compliance

Check against rules defined in conductor/product.md:

| Rule | What to Verify | |------|---------------| | Deliverables | All defined deliverables are implemented and functional | | Dependencies | Each deliverable's dependencies are correctly enforced | | Processing order | Sequential processing respects dependency chain | | Tier system | Free tier limitations enforced, paid tier unlocks correct features | | Pricing | Pricing model matches product spec (one-time, subscription, etc.) | | State rules | State transitions (e.g., lock/unlock, draft/publish) propagate correctly |

markdown
### Product Rules: PASS / FAIL
- Rules checked: [count]
- Violations: [list rule: actual behavior]
- Deliverables functional: [X]/[total]

Pass 2: Feature Correctness

For each feature in the spec, verify it works correctly:

| Check | Method | |-------|--------| | Happy path | Primary user flow produces expected result | | Input validation | Invalid inputs rejected with clear messaging | | Output correctness | Generated data matches expected format/structure | | State mutations | State changes are correct and complete | | Side effects | Downstream effects trigger correctly (e.g., dependency propagation) |

markdown
### Feature Correctness: PASS / FAIL
- Features tested: [count]
- Correct: [count]
- Failures: [describe each]

Pass 3: Edge Cases

| Scenario | What to Verify | |----------|---------------| | Empty state | First-time user with no data | | Boundary values | Max input length, empty inputs, special characters | | Concurrent operations | What happens if user triggers 2 operations at once | | Network failure mid-operation | Partial state handled correctly | | Re-processing | Re-running an operation on existing data prompts confirmation if needed | | All items locked/finalized | UI reflects that no further changes are possible | | Tier limits | Exceeding free tier limit shows upgrade prompt |

markdown
### Edge Cases: PASS / FAIL
- Scenarios checked: [count]
- Unhandled: [list]
- User impact: [describe]

Pass 4: State Transitions

Verify state machine correctness for your project's state model. Example pattern:

| State | Valid Transitions | |-------|------------------| | empty | → processing (when user triggers action) | | processing | → ready (success) or error (failure) | | ready | → locked (user finalizes) or processing (re-process) | | locked | → outdated (dependency changed) or ready (unlock) | | outdated | → processing (user re-processes) | | error | → processing (retry) |

Adapt the state table above to match your project's actual states.

markdown
### State Transitions: PASS / FAIL
- States implemented: [list]
- Invalid transitions possible: [list]
- Missing transitions: [list]

Pass 5: Data Flow

| Check | What to Verify | |-------|---------------| | Input → Processing | User form data correctly feeds into processing pipeline | | Processing → Output | Results stored/displayed correctly | | Output → Persistence | Results saved to store/database | | Cross-component | Data shared correctly between components | | Stale data | No stale renders after state changes |

markdown
### Data Flow: PASS / FAIL
- Flow verified: [input → output]
- Stale data issues: [describe]
- Data loss points: [list]

Pass 6: User Journey Completeness

Walk through the complete user journey for the feature under evaluation. Example structure:

1. User provides input (form, selection, etc.)
2. System processes input
3. User reviews output
4. User can lock/finalize results
5. System handles dependencies between outputs
6. User views all deliverables
7. User can export/download results
8. User can re-process any unlocked item
9. Locked items show "outdated" if dependencies change

Adapt the journey steps above to match your project's actual user flow.

markdown
### User Journey: PASS / FAIL
- Steps completed: [X]/[total]
- Broken at step: [which]
- User experience: [smooth / friction at: describe]

Verdict Template

markdown
## Business Logic Evaluation Report

**Track**: [track-id]
**Evaluator**: eval-business-logic
**Date**: [YYYY-MM-DD]

### Results
| Pass | Status | Issues |
|------|--------|--------|
| 1. Product Rules | PASS/FAIL | [details] |
| 2. Feature Correctness | PASS/FAIL | [details] |
| 3. Edge Cases | PASS/FAIL | [details] |
| 4. State Transitions | PASS/FAIL | [details] |
| 5. Data Flow | PASS/FAIL | [details] |
| 6. User Journey | PASS/FAIL | [details] |

### Verdict: PASS / FAIL
[If FAIL, list specific fix actions for loop-fixer]

Handoff

PASS → Return to loop-execution-evaluator → Conductor marks complete
FAIL → Return to loop-execution-evaluator → Conductor dispatches loop-fixer

Want to see if ibrahim-3d/eval-business-logic would activate on your phrasing? Try it before you install — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/23/2026, 04:28 UTC · 22 cases

Measured on 20 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

+50%

77% vs 27%

Agent turns

1.0 vs 1.0

Tokens

+21%

2,711.15 vs 2,237.4

fail → pass · 12pass → pass · 5fail → fail · 4pass → fail · 1

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 Please run a thorough business logic evaluation on our new subscription and feature packaging pull request. Check our code against the product documentation in `conductor/product.md` along with `spec.md` and `plan.md`. I need an evaluation report that assesses whether tier enforcement and pipeline dependencies adhere to product rules, validates core feature functionality, checks boundary/edge conditions, inspects state machine transitions, verifies end-to-end data flow, and tests user journey completeness. Output a structured report listing the pass/fail breakdown across these six check areas and provide an overall verdict with concrete fix instructions if anything fails.	fail→fail	3,336	26,652	+699%	1	1	0%	248	1,663	+571%	0	0	—
case-02 We've just finished implementing the automated asset generation state machine for track `3d-pipeline-v2`. Please evaluate the business logic by auditing our codebase against `conductor/product.md` and the track specifications. We need a complete assessment covering product rule compliance (dependencies and locking), feature correctness, edge cases like network interruptions or empty inputs, state machine state changes, data flow between components, and user journey steps. Provide a final evaluation report containing a summary table for each check phase, an overall pass/fail status, and a list of required fixes if the audit fails.	fail→fail	25,990	3,166	-88%	1	1	0%	4,531	1,733	-62%	0	0	—
case-03 Can you conduct a business logic evaluation for the deliverable export and workflow locking feature? Use the spec, plan, and product rules in `conductor/product.md` to run the multi-pass verification. We want a report that examines product rule compliance, tests key happy paths and input checks, probes edge cases such as concurrent triggers, audits state transitions, traces data flow from inputs to output storage, and steps through the full user journey. Format the response as an evaluation report with a status grid for each pass and a definitive verdict that specifies remediation actions if any checks fail.	fail→pass	30,388	19,190	-37%	1	1	0%	4,947	4,614	-7%	0	0	—
case-04 Evaluate the payment tier implementation for the track `billing-v1`. Should we check both subscription pricing models and tier unlock logic during the product rules verification step?	fail→pass	13,717	5,693	-58%	1	1	0%	2,128	2,298	+8%	0	0	—
case-05 When auditing core features for track `model-export-v3`, what specific aspects must be verified under Pass 2: Feature Correctness?	fail→fail	15,756	8,996	-43%	1	1	0%	2,373	2,927	+23%	0	0	—
case-06 We are running an edge case assessment on the packaging pipeline for track `zip-generator-v1`. Should network disconnects and concurrent requests be evaluated here, and how are results rendered?	fail→fail	11,266	8,029	-29%	1	1	0%	1,767	2,692	+52%	0	0	—
case-07 Audit the state transitions for a track moving from `processing` to `ready` or `error`. What checks should be documented under Pass 4?	fail→pass	15,886	10,059	-37%	1	1	0%	2,639	3,035	+15%	0	0	—
case-08 Analyze the data pipeline from user form entry through storage persistence for track `data-ingest-v2`. What data flow checks are required?	fail→pass	17,036	9,589	-44%	1	1	0%	2,847	2,962	+4%	0	0	—
case-09 Evaluate the 9-step user journey flow for track `asset-builder-v1`. How should broken journey steps be noted in the evaluation?	fail→pass	14,664	8,080	-45%	1	1	0%	2,299	2,842	+24%	0	0	—
case-10 Before starting a business logic evaluation for track `pricing-matrix-v1`, what project files need to be retrieved from the repo context?	fail→pass	7,879	2,650	-66%	1	1	0%	1,221	1,865	+53%	0	0	—
case-11 Generate a blank Verdict Template structure for a completed business logic evaluation report on track `render-pipeline-v4`.	fail→pass	12,067	2,616	-78%	1	1	0%	2,082	1,923	-8%	0	0	—
case-12 What handoff action occurs when all six evaluation passes achieve a PASS status for track `mesh-decimation-v1`?	fail→pass	9,612	1,673	-83%	1	1	0%	1,450	1,680	+16%	0	0	—
case-13 What handoff action takes place when Pass 3 (Edge Cases) reports unhandled network failure errors during evaluation?	fail→pass	8,600	2,821	-67%	1	1	0%	1,245	1,861	+49%	0	0	—
case-14 When testing deliverable processing order for track `scene-assembly-v1`, what rule check must be enforced in Pass 1?	pass→pass	6,338	1,998	-68%	1	1	0%	925	1,754	+90%	0	0	—
case-15 If a locked deliverable has a upstream dependency change in track `texture-baker-v2`, what state transition behavior must be verified?	pass→pass	12,770	5,269	-59%	1	1	0%	2,093	2,305	+10%	0	0	—
case-16 Under Pass 3 (Edge Cases), how should re-running an operation on existing unlocked data be handled according to standard business rules?	fail→pass	12,341	6,518	-47%	1	1	0%	1,992	2,425	+22%	0	0	—
case-17 When testing input validation under Pass 2 (Feature Correctness) for track `user-tier-v1`, what expectation is set for invalid entries?	pass→fail	6,311	1,768	-72%	1	1	0%	980	1,705	+74%	0	0	—
case-18 What key validation step is performed for multi-component data architecture in Pass 5 (Data Flow)?	fail→pass	8,998	2,274	-75%	1	1	0%	1,484	1,742	+17%	0	0	—
case-19 Which orchestration component dispatches the `eval-business-logic` evaluator when a track implementing state machines or pricing logic finishes?	fail→pass	9,790	2,366	-76%	1	1	0%	1,511	1,745	+15%	0	0	—
case-20 Please write a comprehensive Jest unit test suite with mock functions for testing a JavaScript pricing calculation function.	pass→pass	21,639	24,705	+14%	1	1	0%	4,813	6,805	+41%	0	0	—
case-21 Design a PostgreSQL database schema for tracking user subscription tiers, deliverables, and asset processing states with foreign key constraints.	pass→pass	19,780	15,115	-24%	1	1	0%	4,178	4,472	+7%	0	0	—
case-22 How do I resolve a Git merge conflict between main and feature branches in `conductor/product.md` using command line tools?	pass→pass	9,364	6,279	-33%	1	1	0%	1,774	2,571	+45%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +50 percentage points is the difference between those two pass rates over the 20 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 27.3%
Pass rate with the skill: 77.3%
Cases attempted: 22
Cases counted toward lift: 20
Cases that regressed: 2
Skill version measured: v1
Run completed: 8/23/2026, 04:28 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-03

Prompt

Can you conduct a business logic evaluation for the deliverable export and workflow locking feature? Use the spec, plan, and product rules in `conductor/product.md` to run the multi-pass verification. We want a report that examines product rule compliance, tests key happy paths and input checks, probes edge cases such as concurrent triggers, audits state transitions, traces data flow from inputs to output storage, and steps through the full user journey. Format the response as an evaluation report with a status grid for each pass and a definitive verdict that specifies remediation actions if any checks fail.

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run77

Live evalNeeds ≥20 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+30% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 6 days ago

Other measured skills in the registry, with their headline benchmark lift.