Hopeids

What This Skill Does

Hopeids is a sophisticated, inference-based intrusion detection system (IDS) specifically engineered for the OpenClaw AI ecosystem. Designed by emberdesire, it acts as a gatekeeper between incoming data streams and your AI agent, enforcing non-negotiable security invariants. Unlike standard filters, Hopeids performs a high-fidelity risk assessment on incoming messages, utilizing configurable thresholds to decide whether to allow, warn, or strictly block incoming requests. If a message is deemed malicious, it is immediately quarantined with metadata-only logging—ensuring that raw malicious payloads are never stored, processed, or re-injected into your agent's memory (jasper-recall).

Installation

To install the Hopeids skill, run the following command in your OpenClaw terminal: clawhub install openclaw/skills/skills/emberdesire/hopeids

Once installed, you must add the plugin configuration to your OpenClaw configuration file located at ~/.openclaw/config.json. Ensure you define the agents object with your preferred strictMode and riskThreshold settings to tailor protection to specific agent roles.

Use Cases

• Public-Facing Agents: If you have an agent responding to user input, use Hopeids in strictMode to prevent prompt injection or malicious command execution.
• Enterprise Security: Use the Telegram alert integration to receive real-time notifications for every blocked security threat, allowing administrators to maintain human-in-the-loop oversight.
• Internal Tool Automation: Protect sensitive internal agents by setting lower risk thresholds to catch reconnaissance or probing behavior before it affects your backend systems.

Example Prompts

1. "/approve [threat_id_123]"
2. "/trust [user_id_or_handle]"
3. "/quarantine list --verbose"

Tips & Limitations

• Strict Mode: Always use strictMode: true for agents exposed to untrusted external internet traffic to guarantee the full abort pipeline.
• Performance: While the scan happens before processing, setting very complex threshold policies across many agents may introduce negligible latency.
• Metadata Only: Remember that Hopeids does not store the content of blocked messages for privacy and security reasons. If you need to debug a false positive, you must act before the message is fully purged from volatile memory.
• Human-in-the-loop: Ensure your Telegram API tokens are correctly configured to receive alerts, otherwise, the quarantine system will operate in a 'silent block' state, which may hinder your ability to identify active attack patterns against your infrastructure.