惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
Jina AI
Jina AI
Hugging Face - Blog
Hugging Face - Blog
博客园 - 三生石上(FineUI控件)
博客园 - 【当耐特】
大猫的无限游戏
大猫的无限游戏
IT之家
IT之家
宝玉的分享
宝玉的分享
WordPress大学
WordPress大学
有赞技术团队
有赞技术团队
Apple Machine Learning Research
Apple Machine Learning Research
酷 壳 – CoolShell
酷 壳 – CoolShell
阮一峰的网络日志
阮一峰的网络日志
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
爱范儿
爱范儿
小众软件
小众软件
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
The Cloudflare Blog
S
SegmentFault 最新的问题
博客园 - Franky
博客园_首页
T
Tailwind CSS Blog
雷峰网
雷峰网
罗磊的独立博客

Show HN

Show HN: AI agents for UK GDAD PCF roles and their skills The Two Pillars: Mixer Mode and Meta-Software in the Reorganization of Software Work After AI GitHub - JaiCode08/teleport-env What 1,000+ Harness Experiments Taught Me About Self-Improving Agents Show HN: Liiists, a Markdown-first, iOS and CLI list app SwiperTab – Get this Extension for 🦊 Firefox (en-US) GitHub - kouhxp/fftext: Summarize, explain, fact-check, or translate any text, URL, or file. No GPU. No cloud. One command GitHub - sweetpad-dev/sweetpad: Develop Swift/iOS projects using VSCode GitHub - dogmaticdev/IRON: IRON a.k.a. Intermediate Representation Object Notation is a Interpreter/Database that is used to create Programming Languages. GitHub - sjhalani7/vaen: Package your AI coding harness into a portable .agent file, and share it across repos, teams, & the community without ever having to copy-paste instructions, skills, MCP config, or secrets. Show HN: Gandalf the Grader Show HN: Citadeld – replay any CI failure locally from a single file GitHub - tdortman/cuSBF: High-Performance GPU Super Bloom Filter coral-ai/claude-code-token-xray at main · Coral-Bricks-AI/coral-ai GitHub - ulyssestenn/funes: Funes is a Git-based framework for LLM-managed knowledge work: an AI Librarian ingests raw sources, builds an interlinked Markdown knowledge base, and uses it to produce cited reports, analyses, and other outputs. GitHub - ThatXliner/gah: Git Add Hunk, built for agents to use GitHub - harmont-dev/harmont-cli: Command-line client for the Harmont CI platform GitHub - brooksmcmillin/mcp-authflow: OAuth 2.0 Authorization Server framework for MCP servers GitHub - javaid-codes/audit-supply-chain-agents GitHub - amorey/gochan: A small library of common channel architectures for Go, inspired by Rust GitHub - arifozgun/OpenGem: Free, Open-Source AI API Gateway with Gemini, OpenAI & Anthropic Compatibility in 1 file GitHub - Pranesh950/BioPetals: 🌸 Run BIOxAI models at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading GitHub - cnguyen14/bounty-doctor: Diagnose a GitHub bounty issue before you waste hours: detects honeypot scam repos, AI-bot attempt swarms, and stale contests. Show HN: CoreMCP – MCP Server for On-Prem DBs Show HN: KittyHTML – Render HTML/CSS as an inline image in your terminal GitHub - bingud/filemat: Web-based file manager Show HN: TruthLens – Free multi-signal deepfake image detector GitHub - apexlocal-jz/claude-usage-tray: Windows system-tray app showing your Claude Code rate-limit usage at a glance. Zero deps, ~300 lines of PowerShell. Cross-IDE (works regardless of VS Code, Cursor, plain terminal). Release v0.1.2.1 · kouhxp/yapsnap GitHub - noopolis/moltnet: Self-hostable chat network for AI agents. Pre-built bridges for Claude Code, Codex, and the Claws. Rooms, DMs, history. No Slack bots, no Matrix, no glue code.
GitHub - SawyerHood/omegacode: Code based orchestration f...
sawyerjhood · 2026-06-14 · via Show HN

omegacode

CI

An agent-agnostic implementation of Claude Code's Workflows. omegacode runs JavaScript workflow files that orchestrate fleets of coding agents with a small deterministic DSL — agent() / parallel() / pipeline() / phase() — and the workers are pluggable: the same workflow can drive Claude Code, Codex, OpenCode, and pi in a single run.

Install

npm install -g omegacode
omegacode install-skill

install-skill teaches your agents how to author and run workflows by copying the skill into ~/.claude/skills/ (Claude Code) and ~/.agents/skills/ (Codex and other agents). Pass --claude or --agents to install to just one.

You'll need Node 20+ and at least one worker installed: codex (the default provider), claude, opencode (≥ 1.16.2), and/or pi (≥ 0.79.1, npm i -g @earendil-works/pi-coding-agent). Run omegacode doctor to check — it flags binaries below the minimum versions, which the workers refuse at runtime.

Note on opencode/pi sandboxing: neither CLI can enforce a confined sandbox, so omegacode accepts them only with an explicit sandbox: "danger-full-access" (per call or via --sandbox). The default read-only sandbox is rejected with an error naming the remedy — a deliberate fail-closed choice. Model strings pass through verbatim to the backend (e.g. agent("…", { provider: "opencode", model: "openrouter/anthropic/claude-sonnet-4.5", sandbox: "danger-full-access" })).

Use it

With the skill installed, just ask your agent:

use omegacode to adversarially review this PR with both claude code and codex

It will author a workflow — finders fan out in parallel, a cross-provider skeptic pass tries to refute each finding, a synthesizer merges what survives — then run it and report back. Runs are journaled and resumable, and omegacode serve opens a live dashboard of every agent as it works.

What a workflow looks like

export const meta = { name: "adversarial-review", description: "find bugs, cross-examine them" }
// FINDINGS and VERDICT are plain JSON Schemas, elided here

phase("Find")
const findings = await parallel(
  ["correctness", "security", "performance"].map((lens) => () =>
    agent(`Review the diff through the ${lens} lens. List concrete issues.`, { schema: FINDINGS })),
)

phase("Verify")
return await pipeline(
  findings.filter(Boolean).flatMap((f) => f.issues),
  (issue) => agent(`Try to refute: ${issue.desc}`, { provider: "claude-code", model: "claude-fable-5", schema: VERDICT }),
)

Plain JavaScript, no imports — the DSL is injected. Each agent() spawns a real Codex, Claude Code, OpenCode, or pi agent; omit provider/model to inherit whatever the run was started with (--provider --model, default codex), or pin them per call when you want cross-provider diversity. Provider and model are both-or-neither at every site (per-call, meta defaults, CLI flags): a lone provider: or model: is rejected, so a model meant for one provider can never silently ride another provider's call.

CLI

omegacode run <file.workflow.js | name>   # run a workflow (auto-starts the live viewer)
omegacode serve                           # read-only dashboard over all runs
omegacode run <name> --resume <runId>     # resume — only the changed suffix re-runs
omegacode doctor                          # check codex/claude/opencode/pi availability + versions
omegacode guide                           # print the full authoring guide

run also accepts saved workflow names. Six built-ins ship with the package: deep-research, code-review, and four multi-provider workflows that put the two models' decorrelated errors to work — multi-provider-review (both review the same branch independently, then a synthesis merges both), bake-off (both implement the same task in isolated worktrees, blind cross-provider judges pick a winner), provider-debate (propose → attack → rebut for N rounds, then a judge rules), and second-opinion (both answer cheap; agreement returns merged, disagreement escalates to deep effort and adjudicates). Try omegacode run deep-research --args '"your question"', or omegacode workflows to list them. See omegacode guide for the complete authoring reference.