Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Apex Growth V10: Autonomous Intelligence & Outreach Engine
v10.0.0Automates multi-channel lead acquisition, content broadcasting, and email outreach using engagement analytics and persuasion frameworks for optimized growth...
⭐ 0· 189·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
high confidencePurpose & Capability
The name/description claim autonomous lead acquisition and multi‑channel outreach, and the code/instructions implement exactly that (scraping, API broadcasting, SendGrid dispatch, social posts). This functionality is coherent with the stated purpose. However the registry metadata declares no required environment variables or homepage while the SKILL.md and env-example explicitly request many sensitive credentials (SendGrid, X/Twitter, Reddit, FB, Apollo, PROXY, DISCORD_WEBHOOK). That metadata omission is inconsistent and reduces transparency.
Instruction Scope
SKILL.md grants/requests broad runtime privileges (browser full access, shell exec, fs read/write, gmail/send scope) and instructs the agent to run local Python scripts that: stealthily crawl sites to harvest emails, ingest a persuasion playbook to produce manipulative copy, write files, and conditionally auto‑send outreach. Instructions also direct telemetry/notifications to external endpoints (Discord webhook, social APIs, SendGrid). The playbook explicitly guides social‑engineering messaging. This goes beyond benign automation and contains explicit evasion and manipulation steps.
Install Mechanism
There is no formal install spec, but a manifest lists multiple heavyweight Python dependencies (playwright, playwright‑stealth, tweepy, praw, sendgrid, schedule). Playwright will pull browser binaries at install time. Lack of a controlled install script in the skill bundle reduces visibility into what will be written to disk when dependencies are installed.
Credentials
The runtime requires many sensitive credentials (APOLLO_API_KEY, SENDGRID_API_KEY, X/Twitter keys, Reddit creds, FB_PAGE_TOKEN/ID, DISCORD_WEBHOOK_URL, PROXY_URL, SENDER_EMAIL, AUTO_SEND). While these map to the tool's outreach/scraping capabilities, they are high‑sensitivity secrets and the skill bundle/registry did not declare them in the metadata (required env vars = none). The skill also instructs users to copy a .env with those keys (BYOK). This combination increases risk of credential exposure or misuse—especially if AUTO_SEND is enabled.
Persistence & Privilege
always is false, but the repository provides clock_engine.py and README guidance to run it in the background to create a persistent scheduler that triggers scraping, audits, and broadcasting autonomously. Combined with shell:exec, browser access, network outbound, and SEND capabilities, the skill can be made persistent and autonomous outside the LLM context. That gives it a wide blast radius if misused. (No explicit always:true flag, but persistence is achievable via external process.)
What to consider before installing
This skill is an autonomous, weaponized outreach engine: it scrapes websites (stealth mode), harvests emails, crafts high‑persuasion copy from a playbook, and can auto‑send messages via SendGrid and post to social APIs. Before installing: (1) Do not supply real production credentials—test only in an isolated sandbox. (2) Disable AUTO_SEND (set to False) until you’ve manually audited outputs. (3) Review and consider removing or sandboxing the clock_engine.py scheduler to avoid persistent background execution. (4) Be aware this may violate service Terms of Service (scraping, automated posting, unsolicited outreach) and could expose sensitive credentials; only use in controlled, legal contexts such as red‑team testing. (5) If you need only a single feature (analytics or posting), consider extracting and auditing that component rather than running the full package. If you want, I can list the exact env vars the code reads and produce a safe, minimal checklist for sandboxing and a reduced‑risk configuration.Like a lobster shell, security has layers — review code before you run it.
autonomous-agentvk97e64e84j9je49dwpse50h05982z13elatestvk97e64e84j9je49dwpse50h05982z13elow-frictionvk97e64e84j9je49dwpse50h05982z13emcpvk97e64e84j9je49dwpse50h05982z13eopenclaw-skillvk97e64e84j9je49dwpse50h05982z13eosintvk97e64e84j9je49dwpse50h05982z13eragvk97e64e84j9je49dwpse50h05982z13eself-improvingvk97e64e84j9je49dwpse50h05982z13e
License
MIT-0
Free to use, modify, and redistribute. No attribution required.