indirect-injection-detection
Detects indirect prompt injection and goal hijacking in AI agents by evaluating how they process external content like RAG, documents, and web data.
Discover reusable agent skills, browse implementation details, and find the right skill for your workflow.
139 skills found
Detects indirect prompt injection and goal hijacking in AI agents by evaluating how they process external content like RAG, documents, and web data.
Optimize developer experience for multi-component solutions: standardize onboarding, inner-loop, debugging, and cross-platform setup to eliminate friction and tribal knowledge.
Multi-LLM code review pipeline using consensus-based analysis to detect security, architectural, and quality issues.
Automate GitHub issue triage by analyzing reports against the codebase, verifying technical claims, and providing expert-driven responses to resolve invalid issues.
Cascading goal tracking system connecting 3-year vision to daily tasks. Automates progress calculation, stalled goal detection, and project-to-goal alignment for Obsidian vaults.
Deterministic record-replay debugging using rr. Enables reverse execution, crash trace extraction, and GDB-based root cause analysis for software and binary vulnerabilities.
AI-driven GitHub project management using swarm coordination, automated issue triage, project board synchronization, and intelligent task decomposition for efficient development workflows.
Implement professional GitLab CI/CD pipelines with multi-stage workflows, caching strategies, and Kubernetes deployment patterns for scalable automation.
Intelligent pattern selection for Fabric CLI, automatically choosing from 242+ specialized prompts for threat modeling, data analysis, summarization, and content creation.
Generate scaffolding for custom Minecraft Bedrock packet analyzers. Includes template code, registration guides, and packet capture workflows.
Performs a structured five-stage code review covering requirements, correctness, code quality, testing, and security. Provides actionable, categorized feedback (Blocker/Major/Minor/Nit) to improve PR quality.
MIKE-FIRST v6.0: An enterprise multi-cloud resilience platform for compliance auditing, security intelligence, and zero-downtime cloud migration.