[Skill Submission] kate-top-edit

[everyskill-bot]

New Skill Submission

Skill: kate-top-edit
Submitted by: Willie (as Kate)
Reason: Because Kate is a killer

---

This PR was auto-generated from skills.every.to/submit.

AI security review will run automatically.

[cloudflare-workers-and-pages]

Deploying everyskill with    Cloudflare Pages

Latest commit:	f7cdc2d
Status:	✅  Deploy successful!
Preview URL:	https://8d6448bf.everyskill.pages.dev
Branch Preview URL:	https://skill-kate-top-edit-17709144.everyskill.pages.dev

View logs

[github-actions]

Frontier Model Security Review

Agent	Verdict
Claude Opus 4.6	PASS
GPT-5.2	PASS

Final Verdict: PASS

---

Claude Opus 4.6 Analysis

{"verdict": "PASS", "issues": [], "reasoning": "This skill is a straightforward editorial checklist for top-editing drafts. It defines a clear set of writing issues to scan for (vague openers, unsourced quotes, AI-style writing patterns, jargon, hedging, etc.) and specifies a simple checklist output format. There are no prompt injection attempts, no hidden instructions, no encoded commands, no 'ignore previous' patterns. The skill does not reference any tools (Bash, file access, network calls, etc.) and has no allowed-tools specification, so there is no risk of dangerous tool usage or data exfiltration. The description and instructions are entirely focused on editorial review. The mention of searching the 'Every archive' for relevant pieces (item 6) is a benign editorial suggestion rather than a directive to make external network calls to arbitrary URLs. There are no social engineering or manipulation attempts. This is a clean, well-scoped editing skill."}

GPT-5.2 Analysis

{
  "verdict": "PASS",
  "issues": [],
  "reasoning": "No prompt-injection patterns (e.g., hidden/encoded directives, 'ignore previous' instructions) were found. The skill is a constrained editorial checklist generator with strict output formatting and does not request tool execution, system access, network calls, or credential/file retrieval. There are no instructions that attempt data exfiltration, social engineering, or manipulation. No allowed-tools configuration is present, and nothing implies overly permissive tool use."
}

---

Frontier model review complete. Human approval still required.