Silk Code
Any model, cloud or local
Bring whichever model you trust
One provider layer speaks to every OpenAI-compatible endpoint, with native Ollama<br>support and local-model discovery. Switch models mid-session — or per session — without changing<br>how you work.
DeepSeekQwenKimiGLMMiniMax<br>Cloudflare Workers AIOpenRouter<br>OllamavLLMLM Studio<br>any OpenAI-compatible endpoint
Dashed = runs on your own hardware. --model auto picks whatever<br>you have: a running local server first, then any cloud provider with a key. Onboard anything else<br>with silkcode models add — or pull local models Ollama-style with<br>silkcode models pull qwen2.5-coder.
Quick start
Install once. Run it anywhere.
Requires Python 3.10 or newer (macOS ships 3.9 — grab a current Python from<br>python.org first) and git. Keep Silk Code in its own virtual environment, then add a shell alias<br>so the silkcode command is available from every project without activating the<br>environment first.
01
Install or update Silk Code recommended
# install or update Silk Code<br>if [ -d "$HOME/SilkCode/.git" ]; then<br>git -C "$HOME/SilkCode" pull<br>else<br>git clone https://github.com/RupertCloud/SilkCode.git "$HOME/SilkCode"<br>fi
cd "$HOME/SilkCode"
# create the virtual environment if needed<br>[ -d .venv ] || python3 -m venv .venv
# install or update Silk Code<br>.venv/bin/python -m pip install -e .
# add the global command if it is not already configured<br>grep -q 'alias silkcode=' ~/.zshrc 2>/dev/null || \<br>echo 'alias silkcode="$HOME/SilkCode/.venv/bin/silkcode"' >> ~/.zshrc
source ~/.zshrc
silkcode --help<br>This command block is safe to rerun: it updates an existing checkout, reuses its virtual<br>environment, and avoids adding duplicate aliases. The alias invokes Silk Code directly<br>inside its environment, so you do not need to activate .venv to use it.
02
Connect a model — either side works
# cloud: DeepSeek (or qwen / kimi / glm / minimax / openrouter)<br>export DEEPSEEK_API_KEY=sk-your-key
# or local & private: install Ollama from ollama.com, then<br>silkcode models pull qwen2.5-coder<br>Pick a model with tool-calling support — the agent works by calling tools.<br>qwen2.5-coder is the reliable local choice.
03
Point it at a project
cd ~/Projects/my-project<br>silkcode gui . # browser GUI at 127.0.0.1:8377
silkcode gui ~/Projects/my-project # or pass the project path directly<br>silkcode ~/Projects/my-project # terminal REPL<br>silkcode gui ~/SilkCode opens the Silk Code source repository itself.<br>To work on another codebase, pass that project’s directory instead.
Then just ask: “Build a small Flask API with tests, and make sure the tests<br>pass.” Approve its actions as prompts appear; switch to<br>--mode agent once you trust it.
The harness
Everything between you and the model
Multi-session GUI
Project explorer, streaming chat, agent activity timeline, diff viewer. Open parallel<br>sessions with different models; a busy session keeps working while you use another.
Repository intelligence
Every session starts with a repo map, your SILKCODE.md project rules,<br>accumulated project memory, and installed skills — the model knows the codebase<br>before its first tool call.
Permissions with teeth
Commands are risk-classified. Reads run free, writes and installs ask,<br>rm -rf / push / merge always require approval — unless you explicitly<br>grant them for the session.
Checkpoints & revert
Files are snapshotted before the agent touches them. One click (or<br>/revert) undoes a whole turn.
Verifies its own work
run_tests auto-detects pytest, npm, cargo, go, and flutter — the agent<br>runs your suite and reads the failures before claiming success.
Long sessions that survive
Context compaction trims old tool output and turns as you approach the model’s<br>window — sessions roll on instead of dying, and are resumable from GUI or CLI.
MCP client
Connect any Model Context Protocol server — its tools become agent tools,<br>approval-gated like everything else.
Remote sandboxes
Run the agent’s commands in a disposable container instead of your machine:<br>self-hosted with one command, or on Cloudflare Sandboxes via the bundled Worker.
Built-in evals
silkcode benchmark runs real coding tasks end-to-end per model —<br>with an A/B mode that isolates what the harness itself contributes.
GitHub
Authorize like an app, ship like a team
Sign in with GitHub — install the Silk Code app,<br>approve a code in your browser, done. No tokens to create or paste; short-lived credentials,<br>scoped to the repos you chose, refreshed automatically.
silkcode connect githubsign in (or verify a $GITHUB_TOKEN)<br>/push · ⇧ Pushpush when you say so<br>--auto-pushpush automatically after each turn<br>github_create_pragent opens draft PRs, reads issues, merges on approval
And the harness signs its work. Agent-made commits<br>register Silk Code as co-author — you stay the author, the record stays honest:
Fix login redirect after session refresh
Co-Authored-By: Silk Code<br>X-Silk-Model: deepseek/deepseek-chat<br>X-Silk-Session: 42<br>git log --grep=X-Silk-Model<br>forever...