promptguardDetect prompt injection attacks in text. Returns risk score and detected patterns.
Install via ClawdBot CLI:
clawdbot install mirni/promptguardGrade Fair — based on market validation, documentation quality, package completeness, maintenance status, and authenticity signals.
Sends data to undocumented external endpoint (potential exfiltration)
POST → https://api.greenhelix.net/promptguard/v1/scanCalls external URL not in known-safe list
https://api.greenhelix.net/promptguard/v1/scanAudited Apr 18, 2026 · audit v1.0
Generated May 12, 2026
A customer support chatbot uses Promptguard to scan incoming user messages for prompt injection attacks before processing. This prevents malicious users from tricking the bot into revealing sensitive data or executing unauthorized commands. The risk score helps the system decide to block or flag suspicious inputs for review.
An automated content generation platform scans user-provided prompts for injection patterns before feeding them to a language model. Promptguard detects attempts to override system instructions or extract the model's original prompt, ensuring the output remains safe and within intended boundaries. This is critical for platforms that allow user-customized content generation.
A company building an internal AI agent that processes data from multiple external APIs uses Promptguard to sanitize incoming text. Text from external sources might contain injection attempts aimed at compromising the agent's behavior. By pre-scanning, the agent only processes low-risk inputs, maintaining operational security.
An online learning platform implements Promptguard in its AI tutor to detect students attempting to manipulate the tutor into providing answers or bypassing learning objectives. The scanner identifies injection patterns and adjusts the interaction, preserving the pedagogical integrity and preventing cheating.
A company's automated email assistant uses Promptguard to scan incoming emails for prompt injection before composing responses. This prevents attackers from hijacking the assistant to send malicious replies or extract internal knowledge. The risk score triggers alerts for manual review of high-risk emails.
Provide Promptguard as a hosted API with pricing based on number of API calls or characters scanned. Customers pay per use, making it cost-effective for varying volumes. Revenue scales with adoption.
Offer an annual subscription for enterprises that need to self-host Promptguard for compliance or data privacy reasons. Includes volume discounts and dedicated support. Steady recurring revenue from large clients.
Provide a free tier with limited scans per day and basic pattern detection, then charge for higher limits, advanced pattern detection, and integration support. Attracts small users and upsells to paid plans.
💬 Integration Tip
Integrate Promptguard as a middleware layer before your AI model call; it requires only an HTTP POST request, so you can add it in minutes with minimal code changes.
Scored Jul 12, 2026
Security-first skill vetting for AI agents. Use before installing any skill from ClawdHub, GitHub, or other sources. Checks for red flags, permission scope,...
全面排查企业的经营风险情况,适用于供应商准入尽调、贷前风险筛查、合作伙伴背景调查等场景,全方位预警潜在经营风险,辅助决策者规避合作隐患。
Security scanner for AI agent skills. 9 built-in detection signatures. Identifies secrets, unsafe execution patterns, and prompt injection. Sub-50ms results.
Wallet anti-theft guard. One-click scan for high-risk wallet approvals to protect user assets. Use when a user asks for a wallet security check, wallet healt...
Comprehensive security audit for an agent's full skill stack. Chains scanner, differ, trust-verifier, and health-monitor into a single assessment with priori...
Audit and score OpenClaw AgentSkills against structural compliance, quality standards, and OpenClaw-specific architecture patterns. Produces a 0-100 score wi...