AI Engineering

Memory Bridge browses and compares wiki knowledge by which AI tool originally produced it. Use it to explore what each AI tool knows, run diffs between tools, and surface cross-tool blind spots.

Session Search finds a past agent session by topic and loads its context into the current conversation. Use it to find which session covered a topic, when you last worked on something, or load a transcript.

Skill Creator guides users through creating new skills, modifying existing ones, and measuring skill performance with evals and benchmarks. Use it when you need to draft a skill from scratch, iterate on an existing skill, or optimize skill descriptions for better triggering.

Vault Skill Factory distills mature, curated Obsidian wiki pages into a portable, self-contained agent skill (SKILL.md plus references/) written to a review directory for human inspection before installation. Use it when someone asks to package wiki knowledge as a reusable digital expert.

Wiki Agent ingests specific topics from a chosen AI agent's raw conversation history, searching past sessions for a query and distilling relevant blobs into the wiki. Use it when working in one agent and wanting to pull in how you solved something in another agent's history.

Wiki Context Pack produces token-bounded, citation-ready context slices from an Obsidian vault for downstream agents or tasks. Use it when an agent needs bounded, relevant context from a vault without reading the whole vault.

Skill Creator guides the creation, evaluation, and iteration of high-quality agent skills. Use it to plan, write, score, and refine SKILL.md files and reference materials.

Graft is a CLI that builds and queries a codebase graph, returning file:line references and call edges so AI coding agents can locate, understand, and safely refactor code with minimal token usage.

Code Assist guides implementation of code tasks using test-driven development in an Explore, Plan, Code, Commit workflow. It acts as a technical implementation partner and TDD coach, following existing patterns and avoiding over-engineering.

Prompt-Driven Development turns a rough idea into a detailed design document and implementation plan. It guides iterative requirements clarification, research, design, and planning, producing spec-driven artifacts without writing code.

Ralph Docs helps agents answer questions about Ralph Orchestrator's internals by consulting its published llms.txt documentation map instead of guessing. Use it to debug failures, explain behavior, and scope code changes from authoritative docs.

Ralph Hats creates, inspects, validates, and improves Ralph hat collections for multi-hat agent workflows. Use it when making or refining .ralph/hats/*.yml files, debugging hat routing, explaining event topology, or tuning a Ralph run.