Failure Registry
ReviewAudited by ClawScan on May 10, 2026.
Overview
The skill is mostly a registry search tool, but it pulls changing community content and includes advice for stealth social-media automation that users should review before letting an agent act on it.
Use this skill only if you are comfortable with it cloning a public GitHub registry into /tmp and showing community-submitted debugging advice. Do not let the agent automatically apply suggestions, especially those involving stealth browser automation, social-media posting, or session cookies, without explicit review and approval.
Findings (4)
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.
An agent that follows this advice could automate public social-media actions in ways that violate platform rules, trigger account enforcement, or create reputational risk.
The bundled registry content can be returned as a fix or prevention lesson and explicitly recommends stealth browser automation to evade platform bot-detection.
Switched from 'puppeteer' to 'puppeteer-extra' with StealthPlugin... Always use puppeteer-extra + StealthPlugin for any platform that might detect automation
Treat these entries as historical reports, not automatic instructions. Require explicit user approval and policy/terms review before using stealth automation or posting publicly.
If followed outside the registry context, this could expose account sessions or let an agent mutate a social-media account without sufficient oversight.
A bundled example discusses using session cookies for automated account actions. The skill itself does not request these cookies, but search results may encourage sensitive credential/session handling.
Cookies (auth_token, ct0) injected via page.setCookie()... Cron job runs every 3 hours to post tweets, reply, and like.
Do not provide session cookies or account credentials unless the user explicitly intends that workflow; require approval before any account posting, liking, or replying.
Future registry content may differ from what was reviewed and could include poor or unsafe advice.
The script fetches a mutable remote repository into /tmp at runtime, so search results can change outside the reviewed artifact bundle.
REPO_URL="https://github.com/unleashedbelial/agent-failure-registry"... git pull origin main... git clone "$REPO_URL" "$REPO_DIR"
Prefer pinning to a reviewed commit or clearly showing the repository and revision being searched; treat fetched community content as untrusted.
A misleading or malicious registry entry could influence an agent's debugging decisions or future actions.
The skill intentionally brings community-authored recommendations into the agent's context, where they may be over-trusted if not treated as advisory.
examples/ (curated) and submissions/ (community)... Search Results Include: ... Fix that worked ... Prevention strategies ... Lessons learned
Review returned entries before acting on them, and do not let registry text override user instructions, security policies, or normal approval requirements.