Empryo
previously SoulForge
Code in context.
AI coding with a map of your codebase.
Website · Download · Benchmarks · Changelog · Discussions · Discord
SoulForge is now Empryo. Same symbol-level agent, rebuilt around a desktop app and a faster engine. This repository is Empryo's public home for issues and discussions.
Install
# macOS / Linux
curl -fsSL https://empryo.com/install.sh | bash
# Windows (PowerShell)
irm https://empryo.com/install.ps1 | iex
Official installers and direct downloads are available only from empryo.com/download. Empryo is not distributed through Homebrew, WinGet, or npm.
empryo --set-key anthropic sk-ant-... # or run locally with Ollama — no key required
cd your-project
empryo
Desktop app and prebuilt binaries: empryo.com/download. Runs on macOS, Linux, and Windows. Do not download Empryo binaries from GitHub Releases or third-party package managers.
Why Empryo
Most coding agents search, read whole files and patch strings. They never know what depends on the code they just changed. Empryo maps that first:
- It maps before it reads. On launch, Empryo parses your repo into a live graph: every symbol, import and call site, ranked by how much the code leans on it and how often you change it. Those queries answer in milliseconds and cost no tokens.
- It knows what a change will break. Before an edit, the agent sees what imports a file and what usually changes with it. "What breaks if I touch this?" is answered before the first keystroke.
- It edits by symbol, not by string. 70+ symbol-level operations, batches that all land or all roll back, structural edits in 37 languages, and a typecheck before anything is kept. Whitespace cannot break an edit.
- It treats tokens as money. The graph does the navigating that models usually spend context on, so a task takes fewer reads, fewer steps and a smaller bill.
What's inside
| Code genome | a live dependency graph across 37 languages, ranked by importance and by how often files change together, with millisecond search |
| Symbol-level editing | 70+ operations that all land or all roll back, plus structural edits in 37 languages |
| Multi-agent | agents explore and edit in parallel and share what they read, so cheap models scout and strong models write |
| Task router | ten jobs, any model in any seat, set per tab |
| Time machine | every prompt is a checkpoint. Rewind code and conversation together, to any turn |
| Three surfaces | a desktop app, a full terminal UI, and a headless CLI for scripts and CI, all on one engine |
| LSP + MCP | 576 language servers through Mason, any MCP server, and 13 points where your own scripts can run |
| Free compaction | Empryo shortens a long conversation without calling a model for it, so long sessions stay cheap |
One agent, many brains
Empryo is not one model in a loop. It is a crew, and you assign the seats. Every job takes any model from any of the 31 providers:
brain · spark scout · ember code · explore · verify review · goal review · desloppify · summarize · compact · web search
- Per tab. Each tab carries its own routing: a frontier model writing code in one, a fast cheap one triaging issues in the next, a local model on a private repo in a third.
- Per config. Set defaults globally or per project; override any slot from the tab. Cheap models scout, strong models write, reviewers judge with clean context.
- Custom agents. Define your own with a prompt, a model and a tool policy, then dispatch them alongside the built-ins. One run can mix providers freely.
- It protects your prompt cache. Sub-agents reuse the parent's cached prefix, so context you have already paid for is billed at cache rates instead of full price.
- Costs, itemized. Spend tracked live per model, per sub-agent, per tab, per session and per day. You always know where the tokens went.
Benchmarks
Head to head against pi, on the same models, repositories and tasks:
| Round 1 3 bugs × 3 models | Round 2 5 real bugs · hono / zod / ky | |
|---|---|---|
| Bugs fixed | 8/9 vs 7/9 | 7/10 vs 6/10 |
| Cost | 28% lower — $1.13 vs $1.58 | 23% lower — $7.08 vs $9.19 |
| Time taken | 57% faster — 4m 16s vs 10m | 32% faster — 22m 30s vs 32m 55s |
| Efficiency | 5.7× fewer input tokens — 1.09M vs 6.21M | 28% fewer steps — 274 vs 382 |
Round 2 used real bugs from merged PRs (post-training-cutoff, history scrubbed, regression tests injected after each run). Full methodology and transcripts: empryo.com/benchmarks · reproduce at proxysoul/pi-vs-empryo-bench.
Private by design
Empryo runs entirely on your machine. Bring your own key for Anthropic, OpenAI, Google, Groq, DeepSeek, Bedrock and 25 more, use any OpenAI-compatible endpoint, or run fully local with Ollama or LM Studio. Nothing sits in the middle, no code leaves your machine, and there is no per-seat fee. Free to use.
SoulForge
SoulForge remains available to download and install, and continues to receive fixes for bugs and critical issues. New features and active development have moved to Empryo.
brew tap proxysoul/tap && brew install soulforge
# or
bun install -g @proxysoul/soulforge
This repository
- Issues and Discussions are the home for Empryo bug reports, questions and ideas.
- The SoulForge source remains archived here under its existing license (see
LICENSE).
Sponsors
One API, 200+ models, up to 30% off frontier. Wired in as the llmgateway provider.