Binance Square publishing skill for AI agents. Validate session, publish posts, and check status.

This skill implements Binance Square posting and requires your Binance cookie and CSRF token — treat those as highly sensitive credentials. Before installing or running: 1) Verify the apparent metadata mismatch (the registry summary said no env vars but SKILL.md/skill.json require BINANCE_COOKIE_HEADER and BINANCE_CSRF_TOKEN). 2) Prefer running the skill with a disposable Binance account (not your main trading account) and rotate those cookies/tokens after testing. 3) Do not run the developer tools (scripts/analyze-api.ts, scripts/probe-endpoints.ts) with real credentials unless you understand they will capture request headers/post bodies and may write them to disk (api-analysis.json). 4) If you must grant access to a local browser CDP (BINANCE_CDP_URL), only point it at a dedicated, isolated browser instance (not your daily browser) because the scripts inject cookies and evaluate arbitrary fetches in the page context and could access other open pages or data. 5) Audit scripts/bn-square.mjs and the analyze/probe scripts before use (look for any non-Binance outbound endpoints; the SKILL.md does insist on only contacting *.binance.com, but developer utilities can write reports locally). 6) If you are not comfortable reviewing code, do not provide real credentials — use throwaway credentials or decline installation. If you proceed, monitor for unexpected network activity and rotate credentials afterward.

Type: OpenClaw Skill Name: bn-square-skill Version: 0.2.0 The skill is designed to interact with Binance Square, requiring sensitive credentials like cookies and CSRF tokens. Both the code and documentation (SKILL.md, RULES.md, MESSAGING.md) demonstrate a strong emphasis on security, including input validation via Zod, explicit sanitization of sensitive data from logs and error messages using `sanitizeSensitiveText` (src/utils/errors.ts), and strict domain restrictions for network requests to `*.binance.com`. While the use of Chrome DevTools Protocol (CDP) for browser automation (src/api/browser-client.ts) and downloading images from arbitrary URLs (src/utils/image-upload.ts) are powerful capabilities, they are plausibly necessary for the stated purpose and are implemented with mitigating controls like `escapeForJs` and content type/size checks. There is no evidence of intentional harmful behavior or prompt injection attempts against the agent; instead, the markdown files reinforce security best practices.