mandubian

Facilitates multi-step credential onboarding processes, ensuring secure identity verification and OAuth interactions.

Acts as a red-team agent proposing attack patterns for security detection, enhancing the sentinel's ability to identify threats.

Acts as a cron-driven orchestrator for agent evolution, analyzing sessions and managing the evolution pipeline efficiently.

Conducts static code reviews to ensure artifact correctness, security, and compliance with behavioral contracts without executing code.

Runs artifact test suites in a no-network sandbox, ensuring isolated testing without external dependencies.

Installs durable specialist agents from user requests, enabling automated task management and recurring job execution.

Generates wrapper agents to bridge I/O gaps, enhancing integration between tools and targets for seamless data flow.

Facilitates the end-to-end creation and installation of AI agents, managing design, coding, and deployment processes seamlessly.

Evaluates and decides on the evolution of AI agents based on performance metrics and evidence, enhancing their capabilities.

Analyzes cross-session learnings and scores agent performance to enhance AI evolution through structured insights.

Installs new durable agents into the runtime, ensuring proper validation and workflow management.

Translates natural language governance requests into structured constitutional amendment proposals for effective policy management.

Acts as a front-door agent for ambiguous goals, interpreting requests and routing them to specialists for effective resolution.

Architect agent designs structures and decomposes tasks into manageable sub-tasks, facilitating clear project execution.

Auditor agent analyzes code and outputs for correctness, security, and quality, serving as a promotion gate for agent installations.

This skill isolates root causes of errors and proposes targeted fixes, enhancing debugging and root cause analysis capabilities.

Identifies and recommends the best-matched agents for specific task intents using LLM reasoning and agent capabilities.

Executes lightweight shell and script tasks quickly without creating durable artifacts, ideal for basic command execution.

This skill resolves and packages build-time dependencies into artifact layers for network-isolated environments.

This skill enables autonomous evidence collection and research, providing well-cited outputs and managing uncertainty in findings.