AI & ML

metrillm Skill

Benchmark a local LLM model with MetriLLM. Measures performance (tok/s, TTFT, memory) and quality (reasoning, math, coding, instruction following, structured output, multilingual). Use when the user wants to test, compare, or evaluate a local model.

metrillm Skill

Background context about MetriLLM benchmark tool. Activates when the user asks about local LLM performance, model comparison, hardware fitness, or benchmarking. Provides guidance on using MetriLLM CLI and interpreting results.

cyanheads Skill

File a bug or feature request against this repository (cyanheads/git-mcp-server) using the `gh` CLI. Use for tool logic bugs, git provider misbehavior, transport issues, auth/config problems, or new feature proposals.

hive-intel Skill

Use this skill when the user wants Hive from a terminal — "run a CLI command for…", "show me from the terminal", "pipe this into jq", "give me a shell script", "set up a daily briefing", cron jobs, or local diagnostics like hive doctor. Covers install, auth, domain subcommands, JSON/jq output, and scripting patterns. For agent-side tool calls use the MCP path instead; for app code use hive-build.

hive-intel Bundle

Use this skill when the user wants to install, configure, connect, verify, or debug Hive MCP in an AI client — Claude, ChatGPT/OpenAI, Grok, Cursor, Windsurf, VS Code, Gemini CLI, or Codex — including OAuth browser sign-in, headless API-key fallback, missing tools, 401/auth errors, and hosted-vs-stdio questions. For direct CLI or backend key creation use hive-build-onboarding; for calling Hive from app code use hive-build.

nicholasglazer Skill

Measure retrieval quality on your corpus — Hit@5, MRR, nDCG@10, Precision@5. Thin wrapper around `gnosis-mcp eval` with regression tracking against a saved baseline, plain-English interpretation, and tuning pointers when numbers look off. Use after every ingest or config change.

nicholasglazer Skill

First-time setup wizard for Gnosis MCP. Install, init the database, ingest a docs folder, wire your editor — in that order.

nicholasglazer Skill

Populate the gnosis-mcp knowledge base — from local files, git history, or a crawled website. Handles the full matrix of flags (--force, --prune, --wipe, --embed, --include-crawled) in one place.

nicholasglazer Skill

Search the gnosis-mcp knowledge base. Keyword (default), hybrid semantic+keyword (--semantic), or git commit history (--git). Includes sanity checks and a reranker warning.

nicholasglazer Skill

Verify gnosis-mcp server connectivity, schema integrity, and corpus health. Use when MCP calls fail, return empty, or return unexpected data.

nicholasglazer Skill

Load usage-weighted context from Gnosis MCP. Surfaces most-accessed docs for session startup or topic primers.

paiml Skill

Provides automated refactoring suggestions and complexity reduction strategies using PMAT (Pragmatic AI Labs MCP Agent Toolkit). Use this skill when: - User requests code refactoring, optimization, or improvement - Complexity analysis reveals high-complexity functions (cyclomatic > 10) - Code review identifies maintainability issues - Technical debt needs to be addressed systematically - Preparing legacy code for modernization Supports 25+ languages with data-driven refactoring recommendations based on complexity metrics, mutation testing results, and industry best practices (Fowler's refactoring catalog).

paiml Skill

Analyzes polyglot codebases with multiple programming languages using PMAT (Pragmatic AI Labs MCP Agent Toolkit). Use this skill when: - Working with projects containing multiple programming languages - Assessing cross-language integration patterns and quality - Understanding language distribution and architectural boundaries - Comparing quality metrics across language ecosystems - Identifying language-specific best practices violations Supports 25+ languages including Rust, Python, TypeScript, JavaScript, Go, C++, Java, Ruby, PHP, Swift, Kotlin, C, C#, Scala, Haskell, Elixir, Clojure, Dart, Lua, R, and more. Provides unified quality assessment across heterogeneous codebases.

maxanatsko Bundle

Use when writing or fixing DAX query text, choosing between run_query operations, validating results, testing RLS access for roles, or discovering schema before querying. For diagnosing why a query is slow, use mcp-engine-dax-performance; for wrong values, unexpected blanks, or inflated totals, use mcp-engine-dax-debugging; to persist logic as measures, use mcp-engine-semantic-authoring.

maxanatsko Bundle

Use when starting a SemanticOps MCP session, connecting to a Power BI model, checking the current connection, applying saved preferences, composing bulk or write payloads, or recovering from empty results, stale metadata, or argument validation errors. For task work after the session is healthy, use the matching mcp-engine skill (query, schema-authoring, semantic-authoring, testing-changes, security-governance).

maxanatsko Bundle

Use when a user types onboarding, asks to set up SemanticOps MCP, wants help choosing Free vs Pro, or wants licensing, modes, masking, guardrails, preferences, model safety, tests, reporting, diagnostics, RLS testing, or Enterprise posture tailored to their workflow. For changing security, policy, or masking on an already-configured server, use mcp-engine-security-governance.

maxanatsko Bundle

Use when preparing or assessing a Power BI semantic model for Copilot, Fabric data agents, or natural-language Q&A — clear business terminology, unambiguous metrics, usable date defaults, focused field exposure, descriptions, AI instructions, AI data schema recommendations, verified-answer candidates, or natural-language validation tests. For general modeling quality unrelated to AI consumption, use mcp-engine-model-quality.

maxanatsko Bundle

Use when a Power BI measure or query returns wrong values, unexpected blanks, inflated or duplicated totals, a total row that disagrees with its detail rows, a slicer that has no effect, or numbers that disagree with the source system. For slow-but-correct queries, use mcp-engine-dax-performance; for a whole-model assessment, use mcp-engine-model-quality.

maxanatsko Bundle

Use when a DAX query, measure, or visual is slow, when interpreting analyze timings, Storage Engine / Formula Engine splits, or query plans, when VertiPaq storage size or cardinality drives cost, or when the user wants a tuning pass or before/after benchmark. For wrong values, use mcp-engine-dax-debugging; for writing new queries, use mcp-engine-query; for a whole-model assessment, use mcp-engine-model-quality.

maxanatsko Bundle

Use when creating or running model tests, applying test packs, capturing baselines or snapshots, exporting test results, checking dependency impact before a refactor, or using checkpoints, changesets, undo, redo, and rollback. For assessing overall model quality, use mcp-engine-model-quality; for authoring the changes themselves, use mcp-engine-schema-authoring or mcp-engine-semantic-authoring; for orchestrating a full multi-object refactor around these safety tools, use mcp-engine-refactoring.

maxanatsko Bundle

Use when creating, updating, renaming, or deleting tables, columns, calculated columns, relationships, hierarchies, calendars, or partitions, or changing refresh strategy and incremental refresh policy. For measures, calculation groups, or named expressions, use mcp-engine-semantic-authoring; for RLS roles or perspectives, use mcp-engine-security-governance; for multi-object refactors or renames with downstream consumers, use mcp-engine-refactoring.

maxanatsko Bundle

Use when creating or updating measures, KPIs, calculation groups, DAX UDFs, named expressions, Power Query parameters, or model properties, when deciding between a measure and a calculated column, or when cleaning up naming, display folders, and semantic style. For one-off DAX queries, use mcp-engine-query; for physical tables, columns, and relationships, use mcp-engine-schema-authoring; for consolidating or renaming measures with downstream consumers, use mcp-engine-refactoring.

hive-intel Skill

Use this skill when the user needs a Hive API key or account — "I don't have a key", "how do I sign up", "I lost my API key", "set up Hive for me" — or needs browser sign-in versus headless/CI dashboard setup, key prefixes, or plan limits. Guide key creation without exposing secrets in chat, logs, screenshots, or committed files. After the key exists, route to hive-mcp or hive-cli to wire it up.

donbr Skill

Queries genomic databases (Ensembl, NCBI, HGNC) via MCP tools for gene lookup, variant annotation, orthology, and cross-database ID resolution. Falls back to curl when MCP is unavailable. This skill should be used when the user asks to "annotate variants", "find orthologs", "map gene IDs", "analyze linkage disequilibrium", or mentions gene symbols, ENSG IDs, HGNC identifiers, VEP annotation, LD analysis, HGVS notation, or DNA sequences.

donbr Skill

Queries protein databases (UniProt, STRING, BioGRID) via MCP tools for protein lookups, protein-protein interactions, functional enrichment analysis, and cross-database ID mapping. Falls back to curl when MCP is unavailable. This skill should be used when the user asks to "find protein interactions", "analyze interaction networks", "perform GO enrichment", "map protein IDs", or mentions PPI networks, UniProt accessions, STRING scores, BioGRID interactions, or protein ID conversion between databases.

maxanatsko Bundle

Use when creating or changing RLS roles, role filters, OLS permissions, perspectives, allow/deny/confirm policy rules, PII or numeric masking, or audit logging and evidence export, and when deciding whether a request needs security enforcement or only curation. For first-time guided setup of policies, masking, and guardrails, use mcp-engine-onboarding.

ebal5 Skill

Agent Skill のセキュリティ監査。NVIDIA SkillSpector で静的スキャンし、 検出結果の意図を評価して誤検知を切り分け、baseline 登録か修正かを判断する。 以下の依頼時に使用: - 「スキルを監査」「スキルをスキャン」「skill を security チェック」 - 「このスキル入れて安全？」「インストール前に確認」 - upstream-sync PR のレビュー時、外部スキルを取り込む前 - 「skillspector」「skill audit」「scan skill」「audit skill」

ebal5 Bundle

3 次元モデル (統合強度 × 距離 × 変動性) と均衡結合方程式の定義・スケール・ 解釈帯をまとめた共有参照スキル。用語・式・判定基準を単一ソースとして coupling-design-advisor や将来の coupling-audit / coupling-rebalance が参照する。 発火条件: - 「均衡結合」「結合バランス」「3 次元モデル」の用語定義を確認したいとき - balance スコアの解釈帯を参照したいとき - coupling-* 系スキルの前提として読み込まれる 発火しない: - 設計相談そのもの → coupling-design-advisor - 既存コードの計測 → coupling-audit (未実装)

ebal5 Bundle

使い捨ての Python / shell スクリプトを sandbox 内で実行する前に、 haiku Agent で security pre-review を通して安全性を判定する。 サンドボックスは外部ネット exfil は防ぐが secret 読取や CWD 内の source code poisoning は防げないため、実行前チェックが必要。 以下の場面で使用 (他の skill から sub-skill として呼ばれることもある): - 動作確認のため /tmp/check_*.py 等の一時スクリプトを実行 - 調査 snippet を uv run python や .venv/bin/python で走らせる - サブエージェントが生成したスクリプトを実行する 以下では使用しない: - プロジェクトに commit 済みのスクリプト (scripts/run.py 等) の実行 - allowlist 済み固定 entrypoint (pytest, ruff) の起動 - python -c "print(1+1)" レベルの 1 行 snippet (ただし未知モジュールを import する場合は対象)

ebal5 Skill

日本語の技術文書・書籍原稿の文章規範。整形（一文一行、引用ブロック、脚注、コラム記法）、段落と論証の構成（パラグラフライティング）、論証の厳密さ（ツッコミどころの除去）、読み手の負荷の管理、視点と語り、演出の抑制、LLM っぽい空句の禁止、冗長の排除を定める。日本語で技術書の章、草稿、記事、解説文を書くとき、または推敲・リライトするときに使用する。

devcsde Skill

Use when the user wants to check their OATDA account balance, credit usage, or remaining budget before running an expensive AI call. Returns current balance, total usage, and available balance. Triggers on phrases like "check my OATDA credits", "how much budget do I have left", "verify OATDA balance", "am I out of credits", "check usage".

devcsde Skill

Use when the user wants to compare outputs from multiple LLM models side-by-side in a single request. Sends one prompt to multiple providers (OpenAI, Anthropic, Google, DeepSeek, etc.) in parallel via OATDA's /api/v1/compare endpoint and returns each model's response. Triggers on "compare models", "which model is best", "side-by-side LLM", "benchmark models", "gpt-5 vs claude", "evaluate multiple models".

devcsde Skill

Use when the user wants to generate text using OATDA's unified LLM API. Supports 10+ providers including OpenAI, Anthropic, Google, Deepseek, Mistral, xAI, Alibaba, MiniMax, ZAI, and Moonshot.

ntaksh42 Bundle

マルチエージェント協調の設計と実行を支援する。複数エージェント/サブエージェントを束ねたい・並列で調査や実装を分担させたい・orchestrator-worker や supervisor 構成を組みたい・既存のマルチエージェント構成をレビューしたい時に使う。「マルチエージェントを組みたい」「サブエージェントで並列処理して」「エージェント協調を設計/実行して」等で発火。単一エージェントのループ設計は harness-design に委ねる。

amo-tech-ai Bundle

Expert prompt engineering for Claude 4 models (Sonnet 4.5). Use when crafting prompts, optimizing AI responses, implementing chain-of-thought, or improving prompt clarity and effectiveness. Specializes in Claude-specific techniques and best practices.

victorhueni Skill

Execute git commit with conventional commit message analysis, intelligent staging, and message generation. Use when user asks to commit changes, create a git commit, or mentions "/commit". Supports: (1) Auto-detecting type and scope from changes, (2) Generating conventional commit messages from diff, (3) Interactive commit with optional type/scope/description overrides, (4) Intelligent file staging for logical grouping