Sovereign Daily Digest

Compile and summarize configured news, weather, calendar, tasks, quotes, and more into a structured, daily personal intelligence briefing.

This skill is mostly coherent with its goal of creating a local daily briefing, but check a few things before installing: - Inspect the included script (scripts/digest.sh) yourself — it runs locally and reads files under your home directory (e.g., ~/.openclaw/daily-digest/config.yaml, ~/todo.txt, ~/calendars/*.ics, and any paths you add to the config). - The SKILL.md and README mention using an environment variable DIGEST_EMAIL_PASS for IMAP and optionally the 'gh' CLI for GitHub issues, but the skill metadata does not declare these required credentials/tools. If you enable email or GitHub features, you will need to provide credentials (IMAP password via env var or gh-authenticated CLI). Only provide secrets if you trust the source. - Confirm the source/trustworthiness of the repository before running it (the registry header lists 'source unknown' while README/skill.json reference a GitHub repo). If you plan to install via git clone, verify the repository contents and history on the remote. - If you want extra safety, run the script in a limited environment (container or dedicated user account) until you confirm it behaves as expected. Disable scheduling (cron/launchd/Task Scheduler) until you review and trust the code. If you want, I can scan the full scripts/digest.sh content line-by-line and summarize any specific commands that might access network endpoints or sensitive files.

Type: OpenClaw Skill Name: sovereign-daily-digest Version: 1.0.0 The skill is classified as suspicious due to multiple critical shell injection vulnerabilities. The `SKILL.md` explicitly instructs the AI agent to create a crontab entry using a user-controlled `CRON` string without sanitization, leading to potential Remote Code Execution (RCE) via prompt injection (e.g., `echo "${CRON} ..." | crontab -`). Additionally, the `SKILL.md` instructs the agent to use user-controlled `LOCATION` and `REPO` values from `config.yaml` in `curl` and `gh` commands, respectively, which are not robustly sanitized in `scripts/digest.sh` (e.g., `sed 's/ /+/g'` is insufficient for `LOCATION`), also posing RCE risks. These flaws allow an attacker to execute arbitrary commands by manipulating the configuration or agent prompts, despite the skill's stated benign purpose.