Add app growth investigator skill

[dshipper]

No description provided.

[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 detailed but benign product analytics investigation guide. It contains no prompt injection attempts (no 'ignore previous instructions', no encoded commands, no hidden directives). It does not reference any tool usage — there are no Bash commands, no file system operations, no network calls, no external URLs, and no credential access. The skill consists entirely of analytical frameworks, mental models, funnel definitions, and reporting templates for app growth investigation. The referenced files (references/source-patterns.md, references/funnel-patterns.md, references/app-shapes.md) are internal reference documents whose content is partially included inline and contains only analytical guidance. There are no allowed-tools specified, no dangerous operations, no data exfiltration vectors, and no social engineering content. The skill is safe."}

GPT-5.2 Analysis

{
  "verdict": "PASS",
  "issues": [],
  "reasoning": "No prompt-injection patterns detected (no hidden/encoded directives, no attempts to override system/developer instructions, no 'ignore previous' style content). No tool-usage directives are present at all (no Bash commands, file access, network calls, or requests for credentials), so there is no dangerous tool usage or exfiltration vector in the skill text. Content is domain guidance for analytics/product investigation with benign references to internal markdown docs; it does not attempt social engineering, policy evasion, or manipulation. The interface/default_prompt is consistent with the described purpose and does not grant or request overly permissive tools."
}

---

Frontier model review complete. Human approval still required.