Adverse Event Narrative
AdvisoryAudited by Static analysis on Apr 30, 2026.
Overview
No suspicious patterns detected.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If invoked with vague instructions, the agent could read or write unintended local files while preparing narratives.
The skill exposes broad local read/write/edit and shell capabilities. This matches the local JSON-to-narrative workflow, but users should keep operations scoped.
allowed-tools: [Read, Write, Bash, Edit]
Use explicit input and output paths, review generated files before sharing, and avoid granting broad workspace access unnecessarily.
Users have less external context for who maintains the script or how to verify it beyond the included artifacts.
The package includes runnable code but has limited provenance metadata and no install specification. No suspicious code pattern is evidenced, but provenance is still worth checking.
Source: unknown; Homepage: none; No install spec — this is an instruction-only skill; Code file presence: scripts/main.py
Review the included script before running it, prefer a trusted/internal source, and pin or archive the reviewed version if used in regulated workflows.
Case narratives may contain sensitive health information, and accidental inclusion of identifiers could expose patient data in prompts, outputs, or saved files.
The skill is intended to process medical case details and its own references recognize that direct patient identifiers should be excluded.
Patient Information - Age ... Sex ... Weight and height ... Elements to Exclude - Patient identifiers (names, initials, medical record numbers)
Use de-identified case data, exclude names/initials/medical record numbers, and store generated narratives only in approved secure locations.
Overreliance on generated text could lead to inaccurate or incomplete regulatory or pharmacovigilance documentation.
The skill uses high-trust regulatory wording while also warning that qualified medical judgment is needed for some cases. Users should not treat generated narratives as final authority.
Regulatory-grade narrative generation tool ... suitable for submission to FDA, EMA ... Do NOT use when: Case requires medical judgment or causality assessment
Treat outputs as drafts and require review by qualified pharmacovigilance, medical, and regulatory personnel before submission.