AEGIS
v3.4.0Automated Emergency Geopolitical Intelligence System — real-time threat monitoring and safety alerts for civilians in conflict zones. Use when: (1) setting u...
⭐ 1· 424·4 current·4 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (real-time threat monitoring, briefings, Telegram channel delivery) match the included files and declared requirements: Python scripts for scanning/briefings, a source registry, threat regexes, and optional Telegram credentials. The optional LLM support and NewsAPI key are documented and make sense for improving classification/coverage.
Instruction Scope
SKILL.md instructs the agent to create/read a config at ~/.openclaw/aegis-config.json and to store scan state under ~/.openclaw/aegis-data/. It explicitly says the scanner only contacts URLs in references/source-registry.json plus World Monitor, LiveUAMap and Telegram (if enabled). There is no instruction to read unrelated system files or harvest credentials. Note: if you enable cloud LLM verification or Telegram delivery you will be sending scan content (including location/alerts) to those external services.
Install Mechanism
No install spec; this is instruction-led Python code using standard binaries (python3, curl). There is no download-from-URL install step or package manager install in the skill metadata. The code files are bundled with the skill rather than pulled at runtime.
Credentials
No required environment variables; only optional ones (AEGIS_DATA_DIR, AEGIS_BOT_TOKEN, AEGIS_CHANNEL_ID) are listed and are appropriate for local data directory and Telegram delivery. The config schema documents optional API keys (newsapi) and LLM api_key fields. These optional secrets are justified by the described features, but enabling a cloud LLM or supplying a channel bot token will expose alert content and metadata to those external services.
Persistence & Privilege
Skill does not request always:true and does not modify other skills or system-wide settings in the instructions. It stores state/config under its own ~/.openclaw path, which is proportionate for a scanner/cron-driven skill.
Assessment
This skill appears internally consistent with its stated purpose. Before installing or running it, consider: (1) If you enable Telegram channel delivery, you must provide a bot token and channel ID — those credentials will allow the skill to post messages to that channel. Only give the bot token to channels you control. (2) If you enable LLM verification using a cloud endpoint (OpenAI/OpenRouter/etc.), alert text and related context will be sent to that provider; use a local model (Ollama or self-hosted) if you want to keep data private. (3) Review references/source-registry.json to confirm the external sources the scanner will contact; if you need a narrower footprint, disable specific sources in the config. (4) The skill writes state under ~/.openclaw/aegis-data/ and creates ~/.openclaw/aegis-config.json — run it in a sandbox or with a non-privileged account if you want to limit its filesystem impact. (5) Because this is an operational safety tool, test it with non-critical settings (disable CRITICAL instant posting, enable quiet hours) to validate behavior and avoid accidental alerts. If you need higher assurance, review the contents of scripts/aegis_scanner.py and aegis_channel.py for any unlisted outbound endpoints before enabling networked features.Like a lobster shell, security has layers — review code before you run it.
latestvk97cyw26s4xp6nyhjpq1kr6a0983115d
License
MIT-0
Free to use, modify, and redistribute. No attribution required.