Data Analysis
cibersort-immune-infiltration-analysis
Deconvolve bulk RNA-seq or microarray expression into 22 LM22 immune cell proportions using CIBERSORT. Inputs: expression matrix, sample groups. Outputs: immune fraction table, comparison bar plots, statistical test results.
90100Total Score
Core Capability
94 / 100
Functional Suitability
12 / 12
Reliability
11 / 12
Performance & Context
8 / 8
Agent Usability
15 / 16
Human Usability
7 / 8
Security
11 / 12
Maintainability
12 / 12
Agent-Specific
18 / 20
Medical Task
35 / 35 Passed
91Packaged validation run
5/5
90No-plot explicit-column run
5/5
86Zero-permutation boundary
5/5
89Quantile-normalized run
5/5
88Higher-permutation run
5/5
82Missing case group
5/5
83Corrupted signature matrix
5/5
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 | All quantitative statements in the re-audit were grounded in executed runs and recorded artifacts. |
| Practice Boundaries | PASS | The skill remains strictly non-clinical and does not provide diagnosis or treatment recommendations. |
| Methodological Ground | PASS | The implemented and documented workflow stays within a valid local CIBERSORT-style deconvolution scope with explicit limits. |
| Code Usability | PASS | Canonical, qn=true, perm=0, stress, and failure-reporting paths all executed correctly in the audit environment. |
Core Capability94 / 100 — 8 Categories
Functional Suitability
Functional suitability was evaluated against the skill-auditor rubric; no blocking schema-level issue was recorded for this dimension.
12 / 12
100%
Reliability
Failure metadata and payload preservation are now implemented; final promotion is still copy-based rather than a true transactional rename.
11 / 12
92%
Performance & Context
Performance and context was evaluated against the skill-auditor rubric; no blocking schema-level issue was recorded for this dimension.
8 / 8
100%
Agent Usability
The skill communicates outputs clearly, but still relies on artifact inspection rather than a formal response template.
15 / 16
94%
Human Usability
Malformed inputs still fail fast instead of triggering a clarification flow.
7 / 8
88%
Security
No injection or secret-handling issue was found, but logs intentionally retain file paths and run metadata.
11 / 12
92%
Maintainability
Recording logic is now separated into helper and report modules and covered by regression testing.
12 / 12
100%
Agent-Specific
Scope boundaries, failure-path guidance, and trigger precision are strong after the update.
18 / 20
90%
Core Capability Total94 / 100
Medical TaskExecution Average: 87 / 100 — Assertions: 35/35 Passed
91
Canonical
Packaged validation run
5/5 ✓
90
Variant A
No-plot explicit-column run
5/5 ✓
86
Edge
Zero-permutation boundary
5/5 ✓
89
Variant B
Quantile-normalized run
5/5 ✓
88
Stress
Higher-permutation run
5/5 ✓
82
Scope Boundary
Missing case group
5/5 ✓
83
Adversarial
Corrupted signature matrix
5/5 ✓
91
Canonical✅ Pass
Packaged validation run
Generated the full documented output set, including plots, session metadata, and append-only audit files.
Basic 37/40|Specialized 54/60|Total 91/100
✅A1Required outputs were generated for the audited workflow.
✅A2Input handling and validation behaved as documented.
✅A3No unsupported medical or scientific claim fabrication was detected.
✅A4Execution stayed within the stated Data Analysis skill scope.
✅A5Result artifacts and scoring evidence were sufficient for audit review.
Pass rate: 5 / 5
90
Variant A✅ Pass
No-plot explicit-column run
Explicit sample/group column selection worked and the plot directory remained empty as documented.
Basic 36/40|Specialized 54/60|Total 90/100
✅A1Required outputs were generated for the audited workflow.
✅A2Input handling and validation behaved as documented.
✅A3No unsupported medical or scientific claim fabrication was detected.
✅A4Execution stayed within the stated Data Analysis skill scope.
✅A5Result artifacts and scoring evidence were sufficient for audit review.
Pass rate: 5 / 5
86
Edge✅ Pass
Zero-permutation boundary
The run completed successfully and both CLI output and run_record.txt explicitly explained why empirical P-values were NA.
Basic 35/40|Specialized 51/60|Total 86/100
✅A1Required outputs were generated for the audited workflow.
✅A2Input handling and validation behaved as documented.
✅A3No unsupported medical or scientific claim fabrication was detected.
✅A4Execution stayed within the stated Data Analysis skill scope.
✅A5Result artifacts and scoring evidence were sufficient for audit review.
Pass rate: 5 / 5
89
Variant B✅ Pass
Quantile-normalized run
The qn=true branch executed cleanly in the current container.
Basic 36/40|Specialized 53/60|Total 89/100
✅A1Required outputs were generated for the audited workflow.
✅A2Input handling and validation behaved as documented.
✅A3No unsupported medical or scientific claim fabrication was detected.
✅A4Execution stayed within the stated Data Analysis skill scope.
✅A5Result artifacts and scoring evidence were sufficient for audit review.
Pass rate: 5 / 5
88
Stress✅ Pass
Higher-permutation run
The heavier 50-permutation path completed with stable outputs and no format drift.
Basic 36/40|Specialized 52/60|Total 88/100
✅A1Required outputs were generated for the audited workflow.
✅A2Input handling and validation behaved as documented.
✅A3No unsupported medical or scientific claim fabrication was detected.
✅A4Execution stayed within the stated Data Analysis skill scope.
✅A5Result artifacts and scoring evidence were sufficient for audit review.
Pass rate: 5 / 5
82
Scope Boundary✅ Pass
Missing case group
Expected validation failure with a clear SKILL_INVALID_PARAMETER message and persisted failure artifacts.
Basic 32/40|Specialized 50/60|Total 82/100
✅A1Required outputs were generated for the audited workflow.
✅A2Input handling and validation behaved as documented.
✅A3No unsupported medical or scientific claim fabrication was detected.
✅A4Execution stayed within the stated Data Analysis skill scope.
✅A5Result artifacts and scoring evidence were sufficient for audit review.
Pass rate: 5 / 5
83
Adversarial✅ Pass
Corrupted signature matrix
Expected validation failure for non-finite signature values, with persisted run_record and output_manifest entries.
Basic 33/40|Specialized 50/60|Total 83/100
✅A1Required outputs were generated for the audited workflow.
✅A2Input handling and validation behaved as documented.
✅A3No unsupported medical or scientific claim fabrication was detected.
✅A4Execution stayed within the stated Data Analysis skill scope.
✅A5Result artifacts and scoring evidence were sufficient for audit review.
Pass rate: 5 / 5
Medical Task Total87 / 100
Key Strengths
- Implementation, tests, documentation, and output artifacts are now tightly aligned around both success and failure paths.
- Failure handling is materially stronger: invalid reruns no longer erase prior successful payloads and now leave auditable records.
- The perm=0 boundary is now explicit, documented, and traceable in both CLI logs and run_record output.
- Maintainability improved through modularized recording logic and regression coverage for the payload-preservation behavior.