This project uses the eve framework. Before writing code, read the relevant guide from the installed eve package docs. In most installs, those docs are at node_modules/eve/docs/. In workspaces or local package installs, resolve the installed eve package location first and read its docs/ directory. If package docs are unavailable, use https://eve.dev/docs as a fallback. Before implementing an integration yourself, use eve registry search <query> or eve registry list to discover available integrations. Inspect one with eve registry view <item>,…
Durable personal AI assistant built with Eve and Nuxt. This project deploys its Nuxt frontend and eve agent as peer Vercel services through vercel.ts. Before writing agent code, read the relevant guide in node_modules/eve/dist/docs/public/. The Eve agent calls Nuxt over HTTP: Authenticated with Authorization: Bearer <INTERNALAPISECRET>. See apps/web/server/utils/internal-api.ts. Categories: lib/types/memory.ts. One prose block per category; saves replace the full block. See docs/CUSTOMIZATION.md for details.
This private package is the source of the eve/extensions/code extension for CLI-based coding work. It contributes one patch editing primitive, computer use, an authenticated gh tool, sandbox grep, shared PR-watch primitives, investigation and PR skills, instruction fragments, a read-only worker subagent, Connect-backed authentication hooks, and consumer sandbox helpers. Durable prwatch / prwatch_delete wrappers currently live beside e0's consumer mount because eve workflow directives are application-only. Before writing code, read the installed eve package docs for extensions, hooks, tools, skills, subagents,…
Zero is a pre-1 experiment in building an agent-first programming language. Keep public-facing changes honest about what works today without weakening that positioning. Zero is still being shaped around the needs of agents. Breaking changes are acceptable when they move the language, standard library, compiler, or tooling closer to that goal. Do not preserve legacy behavior by default. Prefer the clearer agent-facing design over compatibility shims, migration layers, or carrying old paths forward. Keep examples, docs, tests, and command contracts…
For project context and architecture, see .github/copilot-instructions.md. USE YOUR BEST JUDGEMENT, but when in doubt, test. If your change could affect the built RPMs, smoke-test before reporting success. See azldev-mock. Do NOT skip testing for changes that affect RPM output. Do NOT tell the user "the build succeeded" without also running the smoke-test. If testing cannot be performed (e.g., the package has no runnable binary, or some other issue), explicitly document why and what was verified instead.
⚠️ Distro-wide config — changes here affect every component build. (See azldev.toml — distro/ is shared across all projects.) Default upstream: Fedora 43 (in azurelinux.distro.toml). Inherited by all components unless overridden via build.defines:
Always use composer commands, not any other package manager. The entire unit test suite (composer run-script test) must pass and exit cleanly before you commit code. Integration tests require a running Elasticsearch instance. Skip them if one is not available. No build step is needed - PHP is interpreted directly. All code in src/ and tests/, and any scripts run by composer, must work on Linux, macOS, and Windows. If a specific action you learned to do better will be…
Role: Act as a principal engineer with 10+ years experience in GPU computing and high-performance numerical computing. Focus ONLY on CRITICAL and HIGH issues. Target: Sub-3% false positive rate. Be direct, concise, minimal. Context: cuML C++ layer provides GPU-accelerated ML algorithm implementations using CUDA, with dependencies on RAFT, RMM, cuVS, libcudacxx, thrust, and CUB.
Role: Act as a principal engineer with 10+ years experience in machine learning systems and Python API design. Focus ONLY on CRITICAL and HIGH issues. Target: Sub-3% false positive rate. Be direct, concise, minimal. Context: cuML Python layer provides scikit-learn compatible APIs for GPU-accelerated ML algorithms, supporting cuDF, pandas, and NumPy inputs.
Do not open a PR unless the human operating you has read the issue, understands the problem, and can explain the proposed fix without your help. Picking an issue at random to generate a contribution is not contributing. PRs opened this way will be closed without review. If a human asks you to "find something to work on" or points you at the issue tracker without a specific problem they already understand, stop and tell them to read an issue…
This file helps AI agents understand the structure, tooling, and conventions of the cloudflare-docs repository so they can make correct, buildable changes. This is the source for developers.cloudflare.com. It is…
.flue/ is the Cloudflare Worker that reviews pull requests for cloudflare/cloudflare-docs. It uses Flue 2.1.0, Cloudflare Workflows, R2, Workers AI, and Hono. app.ts verifies GitHub webhooks and delegates to the…
Use this file for repository-wide rules. Human contributors should start with CONTRIBUTING.md. 1. Resolve the target branch (the PR base, not merely the checked-out feature branch), then use the branch context skill. 2. Read versions from build.sbt and environment.yml. Do not copy version numbers into this shared guide. 3. Use the narrowest relevant repository skill: Scala changes, local setup, and code review. 4. For requests to make one or more issues or PRs "5/5", "200% ready", or merge-ready, use the…
GitHub Agentic Workflows (gh-aw) is a GitHub CLI extension that compiles markdown workflows into GitHub Actions. To keep first-turn context small, only these repository root instruction files should be considered ambient: Everything else should be loaded lazily through skills only when needed. Workflows that declare a source: frontmatter entry (for example source: githubnext/agentic-ops@<ref>) are provenance-managed from an upstream bundle. Use skills only when the task requires specialized guidance. Do not pre-load every skill. All skills are local to this repository…
This repository contains the official Go SDK for the Model Context Protocol (MCP). The SDK is designed to be idiomatic, future-proof, and extensible. The project uses the standard Go toolchain.
These defaults are optimized for AI coding agents (and humans) working on apps that deploy to Vercel. - Treat Vercel Functions as stateless + ephemeral (no durable RAM/FS, no background daemons), use Blob or marketplace integrations for preserving state - Edge Functions (standalone) are deprecated; prefer Vercel Functions - Don't start new projects on Vercel KV/Postgres (both discontinued); use Marketplace Redis/Postgres instead - Store secrets in Vercel Env Variables; not in git or NEXTPUBLIC* - Provision Marketplace native integrations with…
Conventions for AI agents and humans contributing to Mistral Vibe — a Python 3.12+ CLI coding assistant managed with uv. Layout: vibe/core is the engine (agent loop, tools, LLM backends, config); vibe/cli is the Textual TUI; vibe/acp bridges to the Agent Client Protocol; vibe/setup runs first-run wizards. Tests live in tests/ with autouse fixtures in conftest.py and test doubles in tests/stubs/. Before architecture-affecting changes, read the matching ADR. If a change fits the current code but conflicts with ADR direction,…
VercelAGENTS.md
