Content Creator Pro
Security checks across malware telemetry and agentic risk
Overview
The skill is coherent as a content-writing tool, but it explicitly pushes AI-written posts to appear like real human experience and can feed ready-to-publish content into another publishing workflow.
Install only if you are comfortable with an autonomous content workflow. Before allowing anything to publish, require human approval, verify all numbers and personal claims, and review the persistent voice/performance/learning files that guide future posts.
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.
Posts may make readers believe they are seeing genuine personal experience rather than AI-assisted content.
The skill explicitly instructs the agent to make generated content be perceived as real human experience, which can mislead audiences and create reputational or authenticity risk.
every piece of content it produces must pass the stranger test — a person who doesn't know you reads it and thinks "real person, real experience," not "generated."
Require human review before publication, disclose AI assistance where appropriate, and only publish personal claims that the user confirms are true.
The agent could create persuasive but inaccurate personal anecdotes, metrics, or proof points in public-facing content.
The instruction encourages adding concrete experiential details and numbers, but the artifacts do not require those details to come from verified user-provided facts.
STEP 4 — SPECIFICITY PASS Find every general statement. Add a specific number, time, or detail. "Many clients" → "11 of the 14 clients I spoke with last quarter"
Constrain the skill to use only verified facts supplied by the user or clearly mark invented examples as hypothetical.
A bad generated post could move from this skill into another agent’s publishing workflow and affect public accounts or brand reputation.
The skill does not publish directly, but it creates queued content intended for a downstream publishing skill, so unreviewed or inaccurate content can propagate to public channels.
Content is written
to /workspace/content/queue/ for acquisition-master to publish.Add an explicit human approval gate between queue creation and publication, especially for claims, personal stories, financial results, or sensitive topics.
Incorrect, poisoned, or overly sensitive entries in these files could be reused in future posts.
The skill reads persistent voice, calendar, performance, hook, and learning files that can influence future outputs.
read:
- /workspace/content/voice.md
- /workspace/content/calendar.json
- /workspace/content/performance.json
- /workspace/content/hooks.md
- /workspace/.learnings/LEARNINGS.mdReview these persistent files regularly and avoid storing secrets, private customer data, or unverified claims in them.