input classification
PassAudited by ClawScan on May 1, 2026.
Overview
This instruction-only skill is coherently focused on classifying user requests, with the main thing to notice being its described logging/audit and downstream handoff of request text.
This appears safe to install as an instruction-only classification skill. Before using it with sensitive requests, check how your environment implements the described logs, audit trail, queues, and human-review path, and ensure secrets or private content are not retained longer or shared more widely than intended.
Findings (2)
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.
User requests may be stored in logs or audit records, which could include sensitive content if the user provides it.
This indicates the skill expects classification decisions and surrounding user-request context to be recorded persistently.
Logs all classification decisions with full context ... Maintains audit trail for all classification decisions
Use appropriate log minimization, access controls, redaction of secrets, and retention limits before relying on the audit trail.
Clarified request text may be routed to other systems or reviewers as part of the classification workflow.
The integration document describes passing classification output, including clarified user input, through downstream queues and to human review.
Classification | Task Decomposition | Event queue ... Classification | Human Review | Priority queue ... Data Provided: clarified_input
Confirm that downstream queues and review paths have clear access controls, identity boundaries, and data-handling rules.