Agent-to-agent (A2A) communication for Amplifier. The behavior behaviors/a2a.yaml composes a client tool (tool-a2a) and an HTTP server hook (hooks-a2a-server, default port 8222), plus a /a2a plain-language expert mode (setup, status, connect, troubleshoot). 1. Unit + lint (fast, host). The modules import amplifier_core, so a bare pytest won't collect — run the suite with the deps wired in: uv run --no-project --with amplifier-core --with aiohttp --with zeroconf \ --with pytest --with pytest-asyncio \ --with-editable modules/tool-a2a --with-editable modules/hooks-a2a-server \ pytest -q Then…
amplifier-bundle-dot-runner implements the vendored strongdm/attractor nlspec faithfully (see README.md "Spec fidelity" and docs/VISION.md). This file exists to make sure that stays true as the repo evolves. Before any design proposal, issue report, or code change touching engine or spec-adjacent behavior, ask yourself: "Did you check what the strongdm/attractor nlspec has to say about this first?" 1. Check the nlspec first. The canonical, pinned copies live in contracts/external/. Cite the section (e.g. "attractor-spec-canonical.md §5.2") in your issue, PR, or design doc.…
Read docs/VISION.md and the relevant contract in contracts/ before changing product behavior. Draft status is not evidence of approval or working software. Read docs/maintainers.md for the normative maintainer curation policy and suggested evaluation practice. - Commit settled product outputs and durable repo guidance. - Keep temporary design drafts, session plans, agent thinking, return logs, and workflow bookkeeping in the surrounding workspace, not this repository. - Do not copy workspace working documents into the repo as context. - The skill that…
For contributors changing this repo. (For agents using the bundle at runtime, see bundle.md, skills/, and agents/ instead -- this file is the other audience: the one editing the code.) Direct pushes are rejected -- work goes through a PR, and merges are squash only (linear history required). Full ruleset: .github/branch-protection-ruleset.json, maintainer notes in CONTRIBUTING.md. One point worth repeating here: required_approving_review_count is 0, not 1-with-a-bypass, because a bypass actor bypasses the entire ruleset -- including required status checks -- which…
The CoreAI Prototype is a React Router single-page app for exploring agent and design system experiences. It uses local mock data by default and consumes @coreai-ds/react-components from this monorepo. From…
This repo uses a dev-trio loop with subagent delegation: * Planner orchestrates everything. Delegates implementation and audit to subagents. Owns ALL logging, notifications, and user questions. The planner is the only agent that talks to the developer directly. * Implementer receives prompts from planner, audits constraints, implements code, runs builds and tests. Returns results to planner. Does NOT log, notify, or ask questions. * Critic receives implementer output from planner, audits against all binding constraints. Returns verdict to planner. Does…
Before creating or changing HTML reports, visualizations, UI components, color palettes, spacing, typography, elevation, or responsive behavior, load and follow: The canonical upstream skill is: Treat the Glint UI system as the source of truth. Do not invent colors or mix unapproved design systems. Preserve accessibility with labels or symbols so color is never the only signal.
This version has breaking changes — APIs, conventions, and file structure may all differ from your training data. Read the relevant guide in node_modules/next/dist/docs/ (resolved from this file's directory; in monorepos the next package may not be visible from the repo root) before writing any code. Heed deprecation notices. This block is written and re-added by next dev — verify at node_modules/next/dist/server/lib/generate-agent-files.js. Removing it from a diff only re-creates the uncommitted change; committing it with your work keeps the tree…
- Treat code as the source of truth for behavior, accessibility, and data contracts. Figma is the source of truth for approved visual composition. - Keep chart data structured and typed. Do not embed report-specific values in drawing logic. - Prefer compound components and explicit variants over boolean configuration props. - Every interactive visual must support keyboard use, equivalent focus and pointer behavior, reduced motion, responsive layouts, and a semantic data table. - Separate factual data, editorial interpretation, methodology, and…
This folder is the documentation site for Fabricator, the desktop app for building Rayfin apps by chatting with GitHub Copilot. It is published at <https://spatney.github.io/rayfin-fabricator/>. This file is the authoring contract: read it before you add or change a page. Many readers are analysts and makers rather than professional developers, and many are opening Fabricator for the first time. Write so they can follow along without help. Every page must read the same as HTML and as Markdown. Appending .md…
A static, single-page web app (index.html) that generates Statements of Work (SOW) for Microsoft Purview / MIP engagements. No build system or backend — the page is self-contained and served statically (e.g. GitHub Pages). This file is the entry point for humans and coding agents. The verify loop checks the basic HTML document and runs the Playwright browser smoke test. The test serves the static site locally and exercises analytics consent, form entry, SOW generation, encrypted JSON export, reset, and…
Guidance for AI coding agents (and any human author) working in this repository. This file is the single source of truth for the lab's scenario facts ("canon") and its MCAP authoring conventions. Read it before editing any page under docs/. This repository is a Jekyll-based training lab (just-the-docs theme, published to GitHub Pages) in Microsoft's MCAP TechWorkshop format. It is the Module 3 hands-on lab — Cross-Layer Attack Lab (End-to-End) — of the L300/L400 workshop Using AI to Secure the…
This repository distributes skills under plugins/<plugin>/skills/<skill>/. Start with CONTRIBUTING.md. - typescript-project-setup, typescript-build-references, typescript-strict, typescript-update, and typescript-publish-types are implemented. Keep unfinished skills' automatic invocation disabled and their not-implemented behavior explicit unless implementing and evaluating them. - Keep plugin metadata synchronized across root, Claude, and Codex manifests, and keep all four marketplace manifests in sync. - Benchmark fixtures and evaluation tooling are maintained separately from the distributed skills. Include evaluation evidence in PRs; do not invent results. - Run npm run validate…
This repo is documentation. Almost every change is prose that someone will read while they follow along with a terminal open, so the writing matters as much as the accuracy.
You will help the user build out this React-based web app that visualizes data from Power BI semantic models. The app fetches live data via DAX queries, renders charts and grids using Vega-Lite and a built-in DataGrid component, and supports light/dark theming. Your job is to discover the user's data, write correct DAX queries, build React components that fetch and display that data, and validate the result in the browser. Break tasks into independent pieces and delegate them to sub-agents…
Manage AI agents (employees) within a company. Returns all agents in the company. This route does not accept query filters. Unsupported query parameters return 400. Returns agent details including chain of command. Returns the agent record for the currently authenticated agent. Temporarily stops heartbeats for the agent. Resumes heartbeats for a paused agent. Permanently deactivates the agent. Irreversible. Returns a long-lived API key for the agent. Store it securely — the full value is only shown once. Manually triggers a…
MicrosoftAGENTS.md