irb-application-assistant
Assists researchers with Institutional Review Board (IRB) application tasks, including drafting informed consent documents, reviewing research protocols for compliance, generating application forms, and preparing submission checklists. Use when the user mentions IRB, Institutional Review Board, research ethics, human subjects research, protocol review, informed consent, or needs help preparing or reviewing an IRB application or submission.
Veto GatesRequired pass for any deployment consideration
| Dimension | Result | Detail |
|---|---|---|
| Scientific Integrity | PASS | The legacy review did not flag invented scientific claims in the package's writing-oriented output. |
| Practice Boundaries | PASS | Practice boundaries held because the package kept to Assists researchers with Institutional Review Board (IRB) application tasks, including... instead of claiming new evidence. |
| Methodological Ground | PASS | The older review treated the package logic as methodologically aligned with its stated workflow. |
| Code Usability | N/A | The audited output is a narrative or formatting deliverable rather than a code-first scientific workflow. |
Core Capability88 / 100 — 8 Categories
Medical TaskExecution Average: 83.6 / 100 — Assertions: 18/20 Passed
The archived evaluation treated Assists researchers with Institutional Review Board (IRB)... as a clean in-scope run.
Use this skill for academic writing tasks that require explicit... remained well-aligned with the documented contract in the preserved audit.
The archived run for Assists researchers with Institutional Review Board (IRB)... confirmed the helper entrypoint and left the workflow in a stable state.
The archived evaluation treated Packaged executable path(s): scripts/main.py as a clean in-scope run.
This stress case was mostly intact, but the archived review centered its concern on: The output stays within declared skill scope and target objective.
Key Strengths
- Primary routing is Academic Writing with execution mode B
- Static quality score is 88/100 and dynamic average is 83.6/100
- Assertions and command execution outcomes are recorded per input for human review