Agent harnessAI and agent developmentModel Context Protocol
dexto↗
@truffle-ai·TypeScript
Configure and run stateful agent applications with tools, memory, and multiple interfaces.
Project overview
Dexto is an agent harness for defining models, tools, and behavior in YAML rather than wiring every workflow by hand. It includes sessions, persistent memory, tool orchestration, multimodal input, and observability, with a ready-made coding agent as an example. Agents can run locally, in the cloud, or through CLI, web, REST, Discord, and Telegram interfaces.
Generate, refine, and export local architecture diagrams and flowcharts.
Project overview
OpenFlowKit is a local-first diagramming studio for creating architecture diagrams and flowcharts from templates, code, structured imports, screenshots, or natural-language prompts. Diagrams remain visually editable, and the studio can export cinematic MP4 walkthroughs. Users may bring keys for several model providers or run fully offline through Ollama. A companion MCP server lets tools such as Claude Desktop and Cursor create and validate diagrams, while project state is stored locally in the browser without a required account or backend.
Query local files and database tables directly with SQL.
Project overview
EasyDB treats CSV, TSV, text, JSON, Excel, and Parquet files as tables that can be queried directly from a macOS or Windows desktop app without installing a separate database or converting the files first. It can also read MySQL and PostgreSQL tables and combine sources with joins, subqueries, and window functions. Dragging a file into the editor generates SQL, and results can be exported as CSV, TSV, or database-specific SQL.
Collection or directoryData and analyticsClaude Code
dbt-agent-skills↗
@dbt-labs·Python
Guide agents through dbt modeling, testing, semantic layers, operations, and migrations.
Project overview
dbt Agent Skills lets compatible assistants automatically load guidance when a natural-language request matches a dbt workflow. The collection covers model development and tests, MetricFlow semantic layers, dbt Mesh governance, job troubleshooting, command usage, and MCP configuration, with separate skills for one-off migrations to Fusion or across data platforms. Most workflows assume dbt is installed and configured and that the user is working in a project containing `dbt_project.yml`.
Generate prioritized AI opportunity maps and implementation roadmaps for companies.
Project overview
This agent skill helps consultants and business teams turn a company name into an AI opportunity map. It guides a web-search-capable agent through company and industry research, pain-point analysis, scenario discovery, prioritization, and implementation planning, with quick, standard, and deep report modes. The repository includes industry examples, analysis frameworks, and report templates. Research quality depends on the connected agent and its search and reasoning capabilities, and the supporting framework script requires Python 3.8 or later.
Maintain a cross-linked local wiki and capture feedback on individual passages.
Project overview
llm-wiki-skill has an agent compile raw material into a persistent Markdown knowledge base, handling writing, filing, cross-references, logs, and maintenance over time. Companion Obsidian and browser tools let users attach feedback to selected text and track open audits. The repository labels the skill experimental and expects it to evolve.
Collection or directorySoftware developmentClaude Code
macos-app-skills↗
@fayazara·Swift
Give coding agents practical patterns for building and shipping native macOS apps.
Project overview
Install selected skills to help an AI coding agent build, maintain, and release native SwiftUI and AppKit applications. The bundle covers command-line Xcode builds, native window and menu patterns, Sparkle updates, notch overlays, and a GitHub release pipeline with packaging and notarization steps. It targets macOS 14+, Xcode 15+, and Swift 5.9+, while the newest Liquid Glass guidance requires newer Tahoe-era tooling.
Collection or directoryResearch and knowledgeClaude Code
qinyan-academic-skills↗
@LeonChaoX·Python
Add research, paper-writing, literature-search, and grant-proposal skills to AI agents.
Project overview
Qinyan Academic Skills is a broad collection for literature search, scientific writing, reviews, research planning, Chinese national grant applications, bioinformatics, drug discovery, clinical research, machine learning, and data analysis. Researchers can install the full library or deploy a category or individual skill to Claude Code, Cursor, Codex, Gemini CLI, OpenClaw, and other supported tools.
Compare skill-enabled and baseline runs to measure whether an Agent Skill actually helps.
Project overview
agent-skills-eval tests an Agent Skill against the same prompts in two modes: with the skill loaded and without it as a baseline. A judge model grades both outputs against defined assertions, producing JSON and JSONL artifacts plus a static HTML report that shows measurable lift, failures, timing, and tool-call behavior. The project includes a command-line runner and a TypeScript SDK, supports OpenAI-compatible chat backends and custom providers, and can run in CI. Users supply both the target model and the judge model.
Collection or directoryProductivity and officeClaude Code
compass-skills↗
@dongshuyan·Python
Keep long-running agent work aligned across tasks, preferences, and handoffs.
Project overview
COMPASS Skills helps preserve the context that usually disappears across long agent sessions. Its local workflows clarify ambiguous work, maintain a repository task graph, create paste-ready handoffs, and keep an auditable collaboration profile; additional skills turn execution history into new skill designs or upgrade plans. Runtime data stays local by default, but the user profile is stored as unencrypted plaintext and should not contain secrets.
Run repeatable evaluations of whether agents discover and use a skill correctly.
Project overview
Skillgrade scaffolds an eval.yaml for a SKILL.md directory, runs repeated tasks, and measures whether an agent discovers and applies the skill as intended. It supports Gemini, Claude, Codex, ACP, OpenCode, and custom command agents, with deterministic or LLM-rubric graders and CLI or browser reports. Node.js 20+ is required, while Docker, API credentials, or a local command depend on the selected executor.
Pair program remotely with low-latency desktop screen sharing.
Project overview
Hopp lets developers select a teammate and start a low-latency screen-sharing session for remote pair or mob programming without exchanging meeting links. Teams can use the managed service or self-host the backend with Docker Compose. macOS is the stable platform today, while Windows remains alpha and Linux support is planned.
Write Markdown in a local-first WYSIWYG editor and review AI changes before applying them.
Project overview
Markra is a local-first Markdown editor available as both a desktop application and a web editor. Users can switch between a WYSIWYG document surface and source mode, work directly with local files, and invoke inline or document-wide AI actions for rewriting, polishing, continuing, summarizing, or translating. Every AI edit is previewed before it is accepted, rejected, or copied. The desktop version adds broader file management, autosave, local spellcheck, and more image-storage options, while the web version depends on browser permissions and provider CORS support.
Bulk-export Apple Notes while preserving folders, formatting, and attachments.
Project overview
Apple Notes Exporter moves an entire iCloud or On My Mac notes library into formats such as Markdown, PDF, DOCX, JSONL, or ENEX while retaining folder hierarchy, rich content, and attachments. Exports can be run from the macOS interface, a scriptable command-line tool, Apple Shortcuts, or the bundled MCP server, and incremental mode rewrites only new or modified notes. The app requires macOS 11 or later and Full Disk Access. Notes stored in Gmail, Outlook, Yahoo, or other email accounts must first be moved into iCloud or On My Mac.
Run self-hosted agents and long-lived multi-agent workflows.
Project overview
SwarmClaw is a self-hosted runtime and dashboard for operating individual agents, swarms, and orchestrators. It provides durable memory, schedules and heartbeats, delegation, background jobs, MCP tools, and connections to many model APIs and coding-agent CLIs. It is aimed at long-running workflows that need observation and operator control, rather than an agent library embedded in another application.
Compile one MDA source into compatible instruction files for multiple agent runtimes.
Project overview
MDA is a Markdown-based specification and CLI for maintaining agent-facing documents from one source. Authors write a single .mda file and compile it into SKILL.md, AGENTS.md, MCP-SERVER.md, and CLAUDE.md outputs, with JSON Schema validation, typed relationships, dependency metadata, content digests, and signature fields. It is intended for authoring and CI checks rather than as an application runtime dependency. A bundled signature verifier, dependency resolver, and central registry are not yet included.
Route prompts by complexity to control LLM API spending.
Project overview
NadirClaw is a self-hosted OpenAI-compatible proxy that classifies each prompt and routes it to a configured simple, mid-tier, or complex model. It can add fallback chains, spending limits, and optional context compaction while remaining compatible with tools such as Claude Code, Codex, and Cursor. The router runs locally and uses provider keys configured by the user. Its source-available core uses the PolyForm Noncommercial 1.0.0 license, so commercial use requires separate authorization.
Collection or directoryMedia, audio and videoClaude Code
skills↗
@inference-sh·Shell
Call generation, search, and model services from an agent.
Project overview
These skills let an agent use the inference.sh CLI for image and video generation, LLM calls, web search, and social automation. The repository also includes JavaScript and Python SDK guidance, agent UI components, and workflow guides for content and product tasks. Cloud-backed use requires the belt CLI and an authenticated inference.sh account.