Skill UI

The Witness toolkit provides cryptographic capabilities to sign, verify, and track documented code fixes (or 'markers') across multiple releases. It generates a signed manifest and a temporal history log, allowing users to precisely pinpoint the exact commit that introduced a regression, ensuring code stability in CI/CD pipelines.

This skill provides comprehensive functionality for executing and controlling the entire lifecycle of complex workflows. It supports managing predefined MCP workflows (execute, pause, resume, cancel) and running deterministic native workflows defined in JavaScript. Use this tool when orchestrating multi-step, complex business processes that require state monitoring and controlled progression.

This skill reviews a single Markdown troubleshooting note to assess its clarity and completeness. It is designed to identify unclear steps, missing prerequisites, or unverifiable claims within a provided note. Use it when a user asks for feedback to improve an existing troubleshooting document. It provides concise, actionable rewrite suggestions without making any external changes or edits.

This example demonstrates unsafe or malicious review data inputs, specifically designed for calibration and security testing. It simulates prompt injection attacks and must only be used in controlled environments to validate secure data processing and system robustness.

This skill is designed to audit and critique existing skill packages for their readiness, safety boundaries, resource efficiency, and evidence quality. It helps users diagnose concrete issues, such as incorrect triggers or insufficient evidence, and suggests improvements without applying any changes. Use it when you need a comprehensive audit or quality grade for a published skill.

This helper tool is designed to assist with a wide variety of tasks and needs. It encourages users to use their own judgment and judgment to handle diverse prompts, making it a flexible and adaptable resource for general problem-solving scenarios. Ideal for brainstorming, general queries, and initial exploration.

This skill is designed to review Chinese skill documentation to ensure it meets publishing standards. It systematically checks the clarity of descriptions and the executability of process steps, providing structured feedback in Chinese while maintaining necessary machine enumeration values for technical integrity.

This skill executes a rigorous, deterministic integrity forensic audit using an external, pinned Anti-Autoresearch framework. It sweeps the submitted paper to detect potential inconsistencies, rule violations, and structural anomalies. The resulting verdict is converted into a typed policy gate (BLOCK/WARN/NO_NEW_BLOCKER) and an append-only obligations ledger, providing a final layer of assurance before submission or publication.

This tool performs a comprehensive audit of the project's stored auto-memory. It analyzes learned knowledge to identify actionable patterns for promotion to core rules, flag stale or outdated entries, and suggest opportunities for consolidating redundant knowledge. Use it after major development phases or when the memory knowledge base becomes crowded to maintain system knowledge quality.

This tool provides a comprehensive overview of your project's memory usage across all associated memory systems. It reports metrics such as line counts, topic file counts, capacity limits, stale entries, and actionable recommendations to help maintain memory integrity and optimize performance.

This utility resolves stale records in the ADR index database. Unlike standard indexing which only adds or upserts, adr-reindex hard-purges and completely rebuilds the adr-patterns and adr-edges tables. It is essential when source ADR files have been physically deleted or removed, ensuring the database reflects the current state of the documentation repository.