Skills from clawhub.ai/0xnagato

Safe Captcha handling via Browser Agent or 2Captcha API.

Intelligent system orchestration for complex tasks using MCP and Pi Agent.

This skill documents the `Hustle` arbitrage engine for game-market monitoring, automation, and vault-based credential management (`hustle/engine/vault.py`, `.vault/secrets.json`). It instructs running commands like `python3 arc-raiders/monitor.py`, accessing/storing secrets in `.vault/secrets.json`, and references external endpoints such as `https://polygon-rpc.com`.

Dangerous skill: executes local shell commands and performs network operations (e.g., `git clone`, `git push`, `gh pr create`) and writes to user paths such as `~/.claude/skills/`. It converts between `SKILL.md` and `gemini-extension.json`, and handles environment variables like `API_KEY` and `PRETTIER_CONFIG`.

Dangerous skill: executes arbitrary shell commands produced by sub-agents via `<<RUN_COMMAND>>` (executed with `subprocess.run(..., shell=True)`) and allows untrusted `<<WRITE_FILE path="...">>` writes to arbitrary paths. It invokes the `gemini` CLI (`gemini chat`), reads workspace files `task_plan.md`, `findings.md`, `progress.md`, and checks the `GEMINI_PATH` env var.