kesslernity/awesome-copilot-studio-agents

103 paste-ready declarative agents for Microsoft 365 Copilot. No coding. No Azure. Deploy in minutes.

What it solves

Setting up a Microsoft 365 Copilot agent usually requires configuration files, coding, or Azure resources. This project removes those barriers by offering a ready-made collection of instruction blocks. Each block defines a single agent's behaviour — you simply paste it into Copilot Studio (or the built-in Agent Builder) and the agent works immediately. The library covers 17 domains and two industry packs (EPC & Energy and Investment Banking & M&A), so most common business tasks already have a pre-built agent.

How it works

Every agent is a self-contained instruction file. Each file includes a name, description, and a full instruction block. To deploy, you open the agent creation page in Microsoft 365 Copilot, enter the name and description, paste the instruction block into the Instructions field, optionally add knowledge sources (like SharePoint sites), and click Create. The agent then becomes available to @mention in Copilot Chat. No coding, no config files, no Azure needed. The instructions are designed to stay within the 8,000-character Copilot Studio limit and include a built-in quality self-check that runs before every response. They also default to formal British English, support French output, and include a banned vocabulary list to avoid AI-sounding language.

Who it's for

Anyone with a Microsoft 365 Copilot licence (the paid add-on, not the standard M365 plan) who wants to automate common workplace tasks without writing code. The agents target roles like project managers, HR, finance, IT, sales, customer success, and executives. Some agents require optional or required knowledge sources (e.g., a SharePoint site for HR policies), while many work with no external data at all.

Highlights

  • 103 production-ready agents across 17 domains plus two industry packs
  • Paste-ready instruction blocks — no coding or configuration required
  • Covers writing, project management, HR, finance, IT, customer success, sales, strategy, ESG, legal, and more
  • Each agent includes a self-check, banned-vocabulary list, and stays within Copilot Studio's instruction limit
  • Deployment via the built-in Agent Builder in Microsoft 365 Copilot — no separate tool needed

Related

  • Project
    FluidInference/FluidUse

    FluidUse is a Swift library for on‑device form‑filling and typed‑decision making on macOS. It reads UI fields via Accessibility, queries tiny Core ML models (form‑filler, Laya, GLiNER2, Kev, Clef‑Vision, Intern‑Decision, Decision 2.0, Vela 2.0, Short‑Reply) and types or acts on the answers, all locally with sub‑millisecond latency on Apple Silicon.

  • Project
    ARCANGEL0/Mitsu

    Mitsu is an agentic penetration-testing framework that automates authorized engagements from reconnaissance to post-exploitation, using an MCP server to connect a chosen LLM to 165+ red-team tools. It keeps persistent per-session intelligence and offers three modes (Recon, Guided, AutoPwn) to balance automation with operator control.

  • Project
    Kohaku-Lab/KohakuTerrarium

    KohakuTerrarium is a Python framework for building AI agents. It provides a reusable runtime with controllers, tools, sub-agents, session persistence, and composition tools, so you can create custom agents quickly without rebuilding the underlying machinery.

  • Project
    pgrundev/pgx

    PGX is a Postgres-compatible runtime that keeps hundreds or thousands of disposable databases alive in one process, with each database costing ~70 KB when idle and being created in milliseconds from a copy-on-write template — aimed at agent, CI, test, and preview workloads, and explicitly not a production primary (v0.1 is not crash-safe, no C extensions, single process).