Morning Coffee Briefing

Daily morning briefing skill. Reads your TASKS.md and memory files, synthesizes a prioritized day plan, and sends it to you via Telegram so you start the day...

This skill does what it says — read your tasks and send a Telegram briefing — but there are a few red flags you should address before installing: 1) Confirm and add the required environment variables to the registry metadata (TELEGRAM_BOT_TOKEN, TELEGRAM_CHAT_ID, TASKS_FILE_PATH) so reviewers and your agent know what secrets are required. 2) Clarify {MEMORY_DIR} and limit exactly which memory files the skill is allowed to read (avoid open-ended 'any memory files' or globs that could expose sensitive data). 3) Remove or explicitly define the 'pipeline status' checks (what services, what credentials, and what endpoints) — otherwise the agent might attempt network calls or require other tokens. 4) Consider removing Bash/Glob/Grep from allowed tools or restrict file read operations to an explicit whitelist to prevent accidental file enumeration or arbitrary shell execution. 5) Test with a throwaway Telegram bot and a sandboxed TASKS.md/memory directory containing non-sensitive data to confirm behavior. If you cannot get explicit metadata for the memory paths and pipeline checks, treat this skill as higher risk and do not grant it access to real personal or production data.

Type: OpenClaw Skill Name: morning-coffee-briefing Version: 1.0.0 The skill is classified as suspicious due to the broad permissions granted by `allowed-tools: Bash` in `SKILL.md`. While the current use of Bash for a `curl` command to send a Telegram message aligns with the stated purpose, the ability to execute arbitrary shell commands creates a significant prompt injection vulnerability. An attacker could potentially exploit this to trick the agent into executing unintended commands or exfiltrating sensitive data by including it in the Telegram message, even though the skill itself does not explicitly instruct such malicious behavior.