Writing Assistant
Security checks across static analysis, malware telemetry, and agentic risk
Overview
This is a benign instruction-only writing helper, but it asks the agent to keep persistent writing style notes and optionally use reminders or other skills.
Safe to install if you want a persistent writing assistant. Before using it, review what you place in writing-state.md, decide whether you want recurring HEARTBEAT.md reminders, and keep optional inbox or content-calendar integrations limited to content you are comfortable sharing with the writing workflow.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
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.
Personal writing samples, style preferences, audience notes, and feedback may accumulate in workspace files and influence future drafts.
The skill intentionally creates persistent writing memory, including voice preferences and past work references, so future outputs can reuse that context.
Stores writing style notes so every output sounds like you
Keep writing-state.md limited to information you are comfortable reusing later, and periodically review or remove outdated or sensitive notes.
If enabled, the agent may continue surfacing writing reminders or checking content status beyond a single drafting session.
The skill offers optional persistent reminder behavior through HEARTBEAT.md, which can cause the agent to keep checking writing-related state over time.
Add to HEARTBEAT.md (Optional) If you want periodic writing prompts or reminders
Only add the HEARTBEAT.md section if you want recurring reminders, and make sure the check frequency and files reviewed match your preferences.
Email-related context from another workflow may be used to draft replies, but the artifacts do not show automatic sending or hidden account access.
The skill can use outputs from another skill, such as inbox-triage, and turn them into draft replies, though it explicitly requires review before sending.
When inbox-triage flags an email needing a reply, I can draft the reply on request. I never auto-send — I always present the draft for review first.
Keep cross-skill integrations explicit, review generated replies before sending, and avoid sharing inbox content with the writing workflow unless needed.