AI & ML

grtninja Bundle

Build, enrich, distill, and validate local training datasets for the Qwen Training Workbench. Use when dataset export, weighted-study assembly, label distillation, workstation-mode enforcement, or training-input validation change together.

grtninja Bundle

Enforce one-brain routing across VRM Sandbox, PC Control, and MemryX shim with direct LM Studio model authority and typed MX3 support. Use when VRM chat/control behavior must be governed by PC Control without changing endpoint roles.

grtninja Bundle

Validate a mixed local-plus-remote AI stack with privacy-safe evidence, endpoint health checks, inference smokes, and repo lockstep gates. Use when a request spans multiple local services, remote hosts, and heterogeneous inference lanes that must be tested end to end without launching restricted avatar endpoints.

grtninja Bundle

Align the four core local app surfaces under the shared meta-harness contract. Use when PC Control, shim/control-center, STARFRAME app, and avatar-runtime defaults or docs drift on model authority, canonical repo root, continuity, or endpoint expectations and one coordinated pass must repair them together.

grtninja Bundle

Compatibility wrapper for legacy local Comfy orchestration requests in <PRIVATE_REPO_B>. Route new MCP/Comfy operations to repo-b-mcp-comfy-bridge.

megastep Bundle

Use when connecting to Xcode via MCP, using xcrun mcpbridge, or working with ANY Xcode MCP tool (XcodeRead, BuildProject, RunTests, RenderPreview). Covers setup, tool reference, workflow patterns, troubleshooting.

megastep Bundle

Use when designing prompts for LLMs, optimizing model performance, building evaluation frameworks, or implementing advanced prompting techniques like chain-of-thought, few-shot learning, or structured outputs.

grtninja Bundle

Run credit-first local Agent Bridge orchestration in <PRIVATE_REPO_B> with strict read-only validation, bounded indexing, and fail-closed guidance hints. Excludes MCP Comfy diagnostics.

grtninja Bundle

Align desktop autonomy, speech fallback, live research, and chat-route runtime behavior across avatar-facing local apps after model-plane or harness shifts. Use when runtime handoff and UI behavior change together.

dtsong Bundle

Use this skill when performing hardware security analysis for System-on-Chip components — threat modeling, verification scaffolding, compliance mapping, executive briefing, microarchitectural attack analysis, physical side-channel assessment, kernel security analysis, emerging hardware security, or TLA+ formal specification. Routes to the appropriate specialist. Trigger phrases include "threat model my SoC", "run STRIDE analysis", "generate SVA assertions", "compliance check against FIPS", "executive summary of findings", "Spectre analysis for cache", "DPA attack assessment", "kernel hardening review", "PQC hardware review", "TLA+ spec for access control". Do NOT use for software-only security, network security, or web application security.

oaustegard Bundle

Builds and revises a complete skill DIRECTORY — SKILL.md, scripts, references, assets — and packages it. Use for "create a skill for X", "turn this into a skill", "update/improve this skill", "why doesn't my skill trigger", "review this SKILL.md", "package this skill", or when a repeated procedure should become a reusable artifact. Enforces the structure that makes skills work — a concrete procedure rather than a goal statement, an explicit applicability boundary, failure modes with their signals, a runtime verification step, and a description checked against the existing catalogue for confusability. For choosing whether the instruction should be a skill at all rather than project instructions or a prompt, use crafting-instructions. For writing quality inside the prose, use writing-instructions.

oaustegard Bundle

Create video from prompts by overseeing multi-clip AI generation end to end: write a shot list, generate each scene with Gemini Omni Flash (via the Cloudflare AI Gateway), review the results, and assemble them into a finished cut. Use when the user asks to make/generate a video, a short film, an animatic, or a multi-scene clip from a script or idea; when they mention Omni, Veo, text-to-video, or image-to-video; or when acting as the editing/director agent over generated footage. Triggers on 'make a video', 'generate a clip', 'short film', 'video from this script', 'turn this into a video', 'omni', 'veo', 'text to video', 'storyboard to video'. For transcoding/trimming/merging/GIF/subtitles use processing-video; for reading or summarizing existing video content use parsing-video.

megastep Bundle

Use when deploying custom ML models on-device, converting PyTorch models, compressing models, implementing LLM inference, or optimizing CoreML performance. Covers model conversion, compression, stateful models, KV-cache, multi-function models, MLTensor.

grtninja Bundle

Keep local training dataset provenance, merged outputs, and manifest truth aligned across dataset-building repos. Use when teacher distillation, enrichment exports, or study manifests change together.

dtsong Skill

Interactive wizard to craft effective prompts using Claude Code best practices

oaustegard Bundle

Check that a document's claims about code are actually true by reading the prose, the code, and the tests and reporting (or fixing) where they disagree. Use whenever the user wants to verify a README, guide, spec, or docstring still matches the code; whenever they mention documentation drift, doc-code sync, "is this still accurate", stale docs, or keeping docs/tests/code consistent; before publishing or merging a docs change; or as a periodic doc-accuracy sweep. The agent reads the prose's meaning directly — there is no claim-comment DSL to maintain. Pairs with TDD — the test suite is the deterministic behavioral gate, this skill is the semantic prose-vs-reality review.

dtsong Skill

Use when writing formal specifications in TLA+ to verify system properties, defining state variables, configuring TLC model checker, and documenting assumptions and limitations. Covers safety and liveness properties for protocols and concurrent systems. Do not use for security claim enumeration without specification intent (use invariant-analysis).

megastep Bundle

Use when implementing speech-to-text, live transcription, or audio transcription. Covers SpeechAnalyzer (iOS 26+), SpeechTranscriber, volatile/finalized results, AssetInventory model management, audio format handling.

megastep Bundle

Reference — all 20 Xcode MCP tools with parameters, return schemas, and examples

oaustegard Bundle

Generate navigable semantic maps from PDF documents. Extracts section structure via font analysis, then runs LLM extraction per section for claims, symbols, and dependencies — all page-anchored. Produces _MAP.md (progressive disclosure), .symbols.json (definition index), .anchors.json (claim references), and a _USAGE.md snippet for CLAUDE.md. Use when analyzing papers, specs, or legal docs; when asked to "map this document", "index this PDF", "what does this paper say"; or when a coding agent needs grounded reference material from a PDF source. Analogous to tree-sitting but for prose documents.

oaustegard Bundle

Query, filter, and transform Markdown structurally with mq — a jq-like CLI for Markdown. Use to extract headings/sections/code-blocks/links from .md files, build a table of contents, pull code blocks of a given language, slice or reshape LLM prompt/output Markdown, or batch-transform docs. Triggers on "extract sections from this markdown", "get all the code blocks", "jq for markdown", "mq", or any structural query over Markdown that grep/Read can't do cleanly.

oaustegard Bundle

Zero-shot univariate time series forecasting using the Reverso foundation model (NumPy/Numba CPU-only inference). Activate when users provide time series data and request forecasts, predictions, or extrapolations. Supports Reverso Small (550K params). Triggers on "forecast", "predict", "time series", "Reverso", or when tabular data with a temporal dimension needs future-value estimation.

oaustegard Bundle

Analyzes an AI/ML publication — paper, preprint, article, technical blog post — and extracts what an enterprise AI engineer should do about it. Use when someone supplies a URL or document on RAG, embeddings, fine-tuning, prompt engineering, agents, or LLM deployment and asks "review this paper", "what do you make of this", "is this worth adopting", or "summarise the method and its limits". The subject matter must itself be machine learning.

dtsong Skill

Golden dataset creation, automated scoring rubrics, and hallucination detection

dtsong Skill

STRIDE-based security threat analysis for proposed features

dtsong Skill

Design ML workflows — experiment tracking, feature stores, model training, serving, monitoring for drift

dtsong Skill

Use when reviewing the hardware/software security interface to verify hardware security feature enablement, IOMMU/SMMU DMA protection configuration, and HW/SW trust model coherence. Covers boot chain verification, memory protection, and control flow integrity features. Do not use for kernel configuration audit (use kernel-hardening) or isolation boundary analysis (use isolation-review).

dtsong Skill

Chunking strategies, embedding pipelines, vector DB selection, and retrieval optimization

megastep Bundle

Use when wrapping UIKit views/controllers in SwiftUI, embedding SwiftUI in UIKit, or debugging UIKit-SwiftUI interop issues. Covers UIViewRepresentable, UIViewControllerRepresentable, UIHostingController, UIHostingConfiguration, coordinators, lifecycle, state binding, memory management.

megastep Bundle

Xcode MCP workflow patterns — BuildFix loop, TestFix loop, preview verification, window targeting, tool gotchas

oaustegard Bundle

Creates production-ready MCP servers using FastMCP v2. Use when building MCP servers, optimizing tool descriptions for context efficiency, implementing progressive disclosure for multiple capabilities, or packaging servers for distribution.

oaustegard Bundle

Assign items to a CLOSED label vocabulary that is too large to put in a prompt — product taxonomies, category hierarchies, tag vocabularies, routing tables, ICD/SIC-style code lists. A cheap model writes the label it thinks the vocabulary would use, and an embedder snaps that writing onto the nearest legal value, so the schema is never transmitted and the output is always in-vocabulary. Use for "classify these into our taxonomy", "tag these against the existing tag list", "map these queries to categories", "the enum is too big to send", or a Literal/enum that hits a provider cap. NOT for a vocabulary that fits in a prompt — structured output measured 0.701 acc@1 there against this pattern's 0.564. NOT for open-ended labelling with no fixed vocabulary, and not for ranked retrieval over documents (bm25).

oaustegard Bundle

Install and drive Google's Antigravity CLI (`agy`) as a non-interactive sub-agent. Use when orchestrating agy, running Antigravity agents from a script or sandbox, delegating a task to Google's agent harness, or wanting a Gemini-backed peer agent alongside Claude.

oaustegard Bundle

Orchestrates parallel API instances, delegated sub-tasks, and multi-agent workflows with streaming and tool-enabled delegation patterns. Routes by surface — native subagents in Cowork and Claude Code, httpx fan-out on claude.ai — and covers Gemini delegation via the Cloudflare AI Gateway on every surface. Use for parallel analysis, multi-perspective reviews, or complex task decomposition.

oaustegard Bundle

Sentence-level and section-level craft for instructions Claude will actually follow, in any container — project instructions, a standalone prompt, or skill body text. Use when the text exists or is being drafted and the question is quality: "why is Claude ignoring this instruction", "rewrite this prompt so it works", "improve my project instructions", "make this instruction stick", "is this too prescriptive", "how much structure does Sonnet need versus Opus". Covers imperative construction, positive framing, strategic-versus-procedural calibration per model, extended thinking, complexity scaling, and a quality checklist. For deciding WHICH container the instruction belongs in, use crafting-instructions. For building, testing and packaging a whole skill directory, use creating-skill.

oaustegard Bundle

Chooses the right FORMAT for instructions on Claude.ai — project instructions, a skill, or a standalone prompt — and gives the format-specific structure once chosen. Use when the question is which container an instruction belongs in ("should this be a skill or project instructions", "where do I put this", "how do I set up this project", "is this worth a skill"), or when someone has instructions and does not know how to package them. For the writing principles that apply inside any of the three formats, use writing-instructions. For building, testing and packaging a complete skill directory, use creating-skill.