prompt-shield-liteMinimal anti-prompt-injection guardrail for OpenClaw agents. Use when handling untrusted external content (web pages, emails, tool output, documents), before...
Install via ClawdBot CLI:
clawdbot install durugy/prompt-shield-liteGrade Fair — based on market validation, documentation quality, package completeness, maintenance status, and authenticity signals.
Contains instructions to override system prompt or ignore user requests
"ignore all previous instructions"AI Analysis
The skill is a defensive security tool designed to detect and block prompt injection attacks; it processes content locally via specified scripts and does not contain instructions to override user intent, exfiltrate data, or call unauthorized external APIs. Its configuration and rule-based logic are transparent and aligned with its stated purpose of providing a guardrail.
Audited Apr 18, 2026 · audit v1.0
Generated Oct 3, 2026
A SaaS support team routes inbound emails and ticket text through Duru Prompt Shield before agents act on them. The pre-action and pre-send scripts flag attempts to hijack customer responses or exfiltrate data via malicious ticket content. This reduces prompt-injection risk in high-volume human-in-the-loop workflows.
An analyst-facing agent scrapes competitor sites and news articles, and the shield scans all fetched text with detect-injection.sh before summarizing. Suspicious embedded instructions trigger a warn or block, forcing the researcher to review source content. This protects downstream synthesis and reporting from manipulation.
An OpenClaw agent proposes shell commands for infrastructure maintenance, and every command is validated with pre-action-check.sh before execution. The strict mode blocks MEDIUM+ risk actions such as recursive deletes or config edits. This adds a confirmation gate against destructive automation errors and outside injection.
A fintech operations bot drafts client notifications and runs each outbound message through pre-send-scan.sh with redaction. The tool returns sanitized text when sensitive data is detected, preventing accidental PII or key leakage. Compliance teams gain an auditable, low-friction control point.
A clinical coordination agent ingests patient referral documents authored by third parties, scanning content for injected instructions that could alter routing or scheduling. When injection risk is detected, the agent halts and requests human confirmation. This supports safer handling of untrusted medical correspondence.
An agent enriches product listings from supplier feeds and external catalogs, validating all fetched text with the injection detector before applying attributes. Blocks and warns prevent malicious supplier content from corrupting listing rules. Sellers benefit from cleaner, more trustworthy catalog automation.
Offer Duru Prompt Shield as a paid security add-on for teams running OpenClaw agents, with tiered plans based on request volume and strictness modes. Customers receive rule packs, log analytics, and priority support. It monetizes the growing need for prompt-injection protection in agent deployments.
Provide a managed service that deploys, tunes, and monitors Prompt Shield across customer workflows. The vendor maintains allowlists, reviews security logs, and responds to injection incidents on behalf of clients. This suits enterprises lacking in-house agent security expertise.
License the shield as a compliance control for regulated industries requiring guardrails on AI-driven outbound communications and document handling. The package includes audit logging, configurable rule files, and reporting exports. Organizations pay for the assurance and evidence trail it produces.
💬 Integration Tip
Copy .env.example to .env, set PSL_MODE and PSL_ACTOR_ID for your risk profile, then wire the three scripts into your agent's pre-action and pre-send hooks before rollout.
Scored Oct 3, 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,...
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...
GEO Audit — AI Search Visibility Checker for ChatGPT, Perplexity, Claude & Gemini. 29-point GEO readiness checklist: robots.txt AI crawler access, Index...
Senior SecOps engineer skill for application security, vulnerability management, compliance verification, and secure development practices. Runs SAST/DAST sc...