Prism Alerts

Real-time Pump.fun token alerts for Solana traders. New launches, graduations, volume spikes. For trading bots, Discord, Telegram, AI agents.

This skill appears to do what it says (poll a PRISM API and produce alerts), but exercise caution before supplying bot credentials or trusting the default PRISM endpoint. Things to check before installing or running: - Verify the PRISM API: the default URL (strykr-prism.up.railway.app) is a third‑party host with no homepage provided; confirm the operator and trustworthiness. Consider self‑hosting or pointing PRISM_URL at a trusted endpoint. - The package metadata does not declare the TELEGRAM/DISCORD env vars shown in the README. Expect to need your own TELEGRAM_BOT_TOKEN / DISCORD_BOT_TOKEN and channel IDs to send alerts — do not paste tokens into unknown web UIs; run the bot locally or in a controlled environment. - Inspect and run the included scripts locally or inside a sandbox/container. The script only uses curl/jq and writes a dedupe file under /tmp, but network requests go to the PRISM service so review traffic if you are concerned about data leaving your environment. - Prefer creating your own messaging bots and supply only those tokens. If you must use an external Prism provider, verify TLS, ownership, and privacy policy; avoid sharing credentials with unknown operators. If you want a higher assurance verdict, provide: the upstream repository or homepage for the PRISM API and the skill, and confirmation of who operates the strykr-prism endpoint; with that info the assessment can move to benign if the operators are trustworthy.

Type: OpenClaw Skill Name: prism-alerts Version: 1.1.2 The skill is designed to fetch real-time cryptocurrency alerts from the Strykr PRISM API. The `scripts/alerts.sh` file uses `curl` to query `https://strykr-prism.up.railway.app` and `jq` to parse the JSON output, storing temporary state in `/tmp/prism_seen_tokens.txt`. All network and file system operations are directly aligned with the stated purpose of providing token alerts. There is no evidence of data exfiltration, malicious execution, persistence mechanisms, or prompt injection attempts against the agent in `SKILL.md`.