ProxyClaw
v2.5.1ProxyClaw by IPLoop — Residential proxy access for AI agents. Route requests through 2M+ residential IPs across 195+ countries from real Android, Windows, Ma...
⭐ 1· 343·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
medium confidencePurpose & Capability
Name/description (residential proxy for AI agents) aligns with required items: curl binary and a single API key (IPLOOP_API_KEY). Included scripts (fetch.sh, QA runner, support server, docs) are consistent with a proxy and QA/scraping toolkit.
Instruction Scope
SKILL.md explicitly instructs the agent to execute ./fetch.sh to perform user-requested fetches and return results (action skill). That is within scope for a proxy-fetch skill. However: the skill will fetch arbitrary user-provided http/https URLs through an external proxy (including potentially internal or sensitive endpoints if the user requests them). The SKILL.md claims 'Files read: None' and that the API key is never echoed; in practice fetch.sh reads/writes a temp file and reads IPLOOP_API_KEY from the environment. The skill bundle also includes other runnable tools (qa_scraper.py, support_ai.py, Docker instructions) that are not mandated to run by SKILL.md but could be executed manually or by an agent with broad discretion.
Install Mechanism
There is no automatic install spec (instruction-only behavior) so nothing is automatically downloaded or executed on install. The repository references third-party artifacts (Docker image ultronloop2026/iploop-node, planned pip/npm packages) and provides commands to run them; these are optional and not installed by the skill itself. If you choose to run Docker images or download binaries referenced in the docs, that will fetch and run external code and should be treated as higher-risk.
Credentials
Only one required environment variable is declared (IPLOOP_API_KEY) which is appropriate for a proxy gateway. The SKILL.md claims the API key is 'checked for existence only, never echoed or logged' and that it is 'not in URL or command line args visible in ps aux' — the fetch.sh implementation passes the key to curl via --proxy-user. Some systems expose process arguments (including curl args) in process listings; this means the key or derived auth string could be seen on systems that show process arguments. Consider this a modest inconsistency to review if you need to keep the key secret from local users or system processes.
Persistence & Privilege
The skill does not request 'always: true' and does not declare any system-wide configuration changes. It provides instructions and optional commands (e.g., docker run --restart=always) that, if executed by the user, create persistent background agents — but that is a manual action, not automatic behavior of the skill. Autonomous model invocation is allowed by default (disable-model-invocation: false) which simply means the agent can run the fetch action when triggered by user requests.
Assessment
This skill appears to do what it says: run requests through IPLoop residential proxies using IPLOOP_API_KEY and curl. Before installing or using:
- Confirm you trust the IPLoop/ProxyClaw service (iploop.io / proxy.iploop.io). The skill will route any requested HTTP/HTTPS URL through an external proxy — do not ask it to fetch sensitive internal URLs, credentials endpoints, or private admin pages unless you accept that external infrastructure will see those requests.
- The only required secret is IPLOOP_API_KEY. Review fetch.sh and your system policy: passing the key to curl with --proxy-user can expose the auth string in process listings on some systems. If that is a concern, run fetches from an isolated host or modify the script to read the password from a file with restrictive permissions or use a curl option that avoids argument exposure.
- The repo includes optional runnable artifacts (a Docker node image, a QA scraper, and a small support HTTP server). These are not executed automatically by the skill but would fetch/run external code if you choose to run them. Treat any docker run / binary download (ultronloop2026/iploop-node) as higher risk and audit or run in sandboxed environment.
- If you rely on the claim that the skill 'does not read files' or 'never logs the key', verify those properties in your environment: fetch.sh writes to a temporary file and other included scripts may read filesystem resources if run.
If you only need occasional web fetches via a third-party residential proxy and you trust IPLoop, this skill is coherent with that purpose. If you have strict secrets or internal-network rules, review or sandbox the fetch.sh and avoid running the optional node agent or support services.Like a lobster shell, security has layers — review code before you run it.
latestvk97b1vb6acv91kd1hb4kthzwa983t93p
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🌐 Clawdis
Binscurl
EnvIPLOOP_API_KEY
Primary envIPLOOP_API_KEY
