Evidence Insight
pytdc
Therapeutics Data Commons (PyTDC) for AI-ready therapeutic ML datasets and benchmarks; use it when you need standardized dataset loading, meaningful splits (e.g., scaffold/cold-start), and consistent evaluation for ADME/Toxicity/DTI/DDI or molecular optimization.
87100Total Score
Core Capability
87 / 100
Functional Suitability
11 / 12
Reliability
10 / 12
Performance & Context
8 / 8
Agent Usability
14 / 16
Human Usability
8 / 8
Security
9 / 12
Maintainability
10 / 12
Agent-Specific
17 / 20
Medical Task
20 / 20 Passed
92You need curated, AI-ready datasets for drug discovery tasks (e.g., ADME, toxicity, bioactivity, DTI/DDI)
4/4
88You want standardized benchmarks with consistent evaluation protocols (including multi-seed benchmark groups)
4/4
86Dataset access by task type
4/4
86Single-instance prediction: ADME, Toxicity (Tox), HTS, QM, and more
4/4
86End-to-end case for Dataset access by task type
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
PASSResearch Veto✅ PASS — Applicable
| Dimension | Result | Detail |
|---|---|---|
| Scientific Integrity | PASS | The archived evaluation kept the skill tied to retrieved records or indexed source material rather than invented scientific claims. |
| Practice Boundaries | PASS | The legacy review kept this workflow on the evidence-access side of the boundary, not the advice-giving side. |
| Methodological Ground | PASS | The legacy audit preserved a method-grounded interpretation of the Therapeutics Data Commons (PyTDC) for AI-ready therapeutic ML datasets and benchmarks; use it when you need standardized dataset loading, meaningful splits (e.g., scaffold/cold-start), and consistent evaluation for ADME/Toxicity/DTI/DDI or molecular optimization workflow. |
| Code Usability | PASS | Code usability passed because the search or lookup workflow still exposed a usable entrypoint and output expectation. |
Core Capability87 / 100 — 8 Categories
Functional Suitability
Related legacy finding for pytdc: Improve stress-case output rigor. Stress and boundary scenarios show weaker consistency
11 / 12
92%
Reliability
Reliability was softened by the legacy issue 'Improve stress-case output rigor'. Stress and boundary scenarios show weaker consistency
10 / 12
83%
Performance & Context
Performance context reached full score in the archived evaluation.
8 / 8
100%
Agent Usability
The legacy audit deducted points for pytdc in agent usability.
14 / 16
88%
Human Usability
Human usability reached full score in the archived evaluation.
8 / 8
100%
Security
The legacy audit deducted points for pytdc in security.
9 / 12
75%
Maintainability
The legacy audit deducted points for pytdc in maintainability.
10 / 12
83%
Agent-Specific
Related legacy finding for pytdc: Improve stress-case output rigor. Stress and boundary scenarios show weaker consistency
17 / 20
85%
Core Capability Total87 / 100
Medical TaskExecution Average: 87.6 / 100 — Assertions: 20/20 Passed
92
Canonical
You need curated, AI-ready datasets for drug discovery tasks (e.g., ADME, toxicity, bioactivity, DTI/DDI)
4/4 ✓
88
Variant A
You want standardized benchmarks with consistent evaluation protocols (including multi-seed benchmark groups)
4/4 ✓
86
Edge
Dataset access by task type
4/4 ✓
86
Variant B
Single-instance prediction: ADME, Toxicity (Tox), HTS, QM, and more
4/4 ✓
86
Stress
End-to-end case for Dataset access by task type
4/4 ✓
92
Canonical✅ Pass
You need curated, AI-ready datasets for drug discovery tasks (e.g., ADME, toxicity, bioactivity, DTI/DDI)
You need curated, AI-ready datasets for drug discovery tasks (e.g.,... was evaluated as a bounded documentation path, not as a runnable script workflow.
Basic 36/40|Specialized 56/60|Total 92/100
✅A1The pytdc output structure matches the documented deliverable
✅A2The instruction path remains actionable for the documented case
✅A3The output stays fully within the documented skill boundary
✅A4The response quality is acceptable for the documented path
Pass rate: 4 / 4
88
Variant A✅ Pass
You want standardized benchmarks with consistent evaluation protocols (including multi-seed benchmark groups)
The archived run for You want standardized benchmarks with consistent evaluation... remained guidance-driven rather than command-driven.
Basic 34/40|Specialized 54/60|Total 88/100
✅A1The pytdc output structure matches the documented deliverable
✅A2The instruction path remains actionable for the documented case
✅A3The output stays fully within the documented skill boundary
✅A4The response quality is acceptable for the documented path
Pass rate: 4 / 4
86
Edge✅ Pass
Dataset access by task type
This edge case stayed inside the documented workflow and remained instruction-led.
Basic 33/40|Specialized 53/60|Total 86/100
✅A1The pytdc output structure matches the documented deliverable
✅A2The instruction path remains actionable for the documented case
✅A3The output stays fully within the documented skill boundary
✅A4The response quality is acceptable for the documented path
Pass rate: 4 / 4
86
Variant B✅ Pass
Single-instance prediction: ADME, Toxicity (Tox), HTS, QM, and more
Single-instance prediction: ADME, Toxicity (Tox), HTS, QM, and more was evaluated as a bounded documentation path, not as a runnable script workflow.
Basic 32/40|Specialized 54/60|Total 86/100
✅A1The pytdc output structure matches the documented deliverable
✅A2The instruction path remains actionable for the documented case
✅A3The output stays fully within the documented skill boundary
✅A4The response quality is acceptable for the documented path
Pass rate: 4 / 4
86
Stress✅ Pass
End-to-end case for Dataset access by task type
The archived run for End-to-end case for Dataset access by task type remained guidance-driven rather than command-driven.
Basic 29/40|Specialized 57/60|Total 86/100
✅A1The pytdc output structure matches the documented deliverable
✅A2The instruction path remains actionable for the documented case
✅A3The output stays fully within the documented skill boundary
✅A4The response quality is acceptable for the documented path
Pass rate: 4 / 4
Medical Task Total87.6 / 100
Key Strengths
- Primary routing is Evidence Insight with execution mode B
- Static quality score is 87/100 and dynamic average is 79.6/100
- Assertions and command execution outcomes are recorded per input for human review
- Execution verification summary: Script verification 0/3; adjustment=0. benchmark_evaluation.py: rc=1; load_and_split_data.py: rc=1; molecular_generation.py: rc=1