morning-briefing

Aggregates weather, game updates, and concert data into a daily markdown briefing. Triggered by 'morning briefing', 'daily briefing', 'run my briefing', 'wha...

This skill appears to do exactly what it says: combine weather from wttr.in and JSON files produced by other skills into a markdown briefing. Before installing or running it, review the following: (1) Inspect the config at ~/.openclaw/config/morning-briefing.json — ensure data_path entries point only to trusted JSON files (do not point to system or secret files). (2) The package as provided does not include assets/templates — templates must come from other skills or you must supply them; a malicious template could alter output, so only use templates you trust. (3) The agent is instructed to present the generated markdown verbatim — any sensitive content present in the aggregated sources will be revealed exactly as-is. (4) The scripts are plain shell; if you plan to run them, review scripts/init_config.sh and scripts/morning_briefing.sh yourself or run them in a safe environment first. If you want to be extra cautious, run the briefing script after setting weather.enabled=false and disabling untrusted sources so it only includes data from known, safe files.

Type: OpenClaw Skill Name: morning-briefing-composer Version: 1.0.1 The skill contains aggressive prompt-injection instructions in SKILL.md and scripts/morning_briefing.sh that explicitly command the AI agent to bypass its standard behavior (summarizing, paraphrasing, or editorializing) and output the generated content verbatim. While the script's primary function is to aggregate data from wttr.in and local JSON files using jq, these instructions are designed to override the agent's safety or processing filters. Furthermore, the script resolves file paths for data sources based on a configuration file, which could potentially be used to read sensitive local files if the configuration is tampered with.