Digitorn
Digitorn
← Back to Learn
Migration guide

From Aider to Digitorn

From a terminal coding assistant to a YAML coding agent.

Aider is a great terminal coding assistant. It pairs with you on a repo, reads files, edits them, and stages git commits. The pattern is excellent, the integration story is the limitation: Aider is a CLI, not a runtime. Digitorn ports the same pattern to a YAML agent that lives anywhere your runtime lives.

Aider is glued to a single user at a single terminal. If you want the same coding assistant inside Slack, on PRs, on a web UI, or running unattended on a cron, you have to leave Aider behind. Digitorn gives you the file-edit primitives, the diff guard, the read-before-edit rule, and the same coding loop, served from a daemon that channels can plug into.

Concept by concept

Every Aider primitive maps to a Digitorn equivalent. Where the mapping is not 1-to-1, the notes call out what changed.

AiderDigitornNotes
/add filefilesystem.readAdding a file is just reading it. The runtime tracks read files for the read-before-edit guard.
edit / diff loopfilesystem.edit (fuzzy match + lint)Edits use a 6-strategy fuzzy matcher. The result includes lint output for instant self-correction.
/run (shell)shell.bashSandboxed Git Bash on Windows, native shell elsewhere. Output goes back to the model as a tool result.
/testbehavior rule (test_after_changes)The behavior engine reminds the agent to run tests when source files change.
git commit on editshell.bash + workspace.commitWorkspace commits are a single endpoint, the agent decides when to call it.
model settingsagents[].brainSwitch model per agent. Coordinator can use a strong model and helpers a faster one.

Side-by-side code

Real apps in both stacks. The Digitorn version is what you would commit to a repo, no scaffolding hidden offscreen.

Terminal coding session vs YAML coding agent

Aider
agent.py
1# Aider, run from the repo:2$ aider --model claude-sonnet-4 src/api.ts src/utils.ts3> add a retry wrapper around the fetch helper, then run the tests45# Aider edits the files, runs npm test, commits if green.6# Works only in your terminal, not in Slack or as a service.
Digitorn
app.yaml
1app:2  app_id: coding-helper3  name: "Coding helper"45execution:6  mode: conversation78modules:9  filesystem: {}10  shell: {}11  behavior:12    config:13      profile: coding   # 14 built-in coding rules1415agents:16  - id: coder17    modules: [filesystem, shell]18    brain: { model: claude-sonnet-4-6, credential: anthropic_main }19    system_prompt: |20      You write code. Read files before editing them. Run tests after21      every change. Use shell.bash for git commands.

The same coding loop as Aider, but the agent now lives in the daemon. Mount the same agent under a Slack channel, a webhook, or a cron, the loop does not change.

What bites people

Subtle differences that look the same on paper and break on first run. Read these before you start porting.

read_before_edit is enforced, not advised

The behavior engine blocks an edit if the file was not read in the same session. Aider has the same idea, Digitorn enforces it at runtime.

Repo map is implicit

Aider builds a repo map at startup. Digitorn relies on filesystem.glob and filesystem.grep instead, the agent maps as it goes. Faster, but the agent has to learn to discover.

Voice mode

Aider has a voice mode that Digitorn does not match yet. If voice is non-negotiable, keep Aider for the voice loop and back it with Digitorn for everything else.

Where Aider still wins
  • Hyper-fast repo map and code completion in a single terminal
  • Built-in voice mode for hands-free coding
  • First-party experience with hosted Anthropic and OpenAI models
Try it in 5 minutes

Install Digitorn and port one of your Aider agents

# 1. install runtime
curl -sSL https://digitorn.ai/install | sh

# 2. copy the YAML above into a file
mkdir -p ~/.digitorn/apps/from-aider
# paste the YAML into app.yaml

# 3. deploy and chat
digitorn deploy from-aider
digitorn dev chat from-aider
Read the deep dive

10 apps you can ship in 50 lines of YAML

Read article
Newsletter

Get the next post in your inbox.

Engineering notes from the Digitorn team. No marketing, no launch announcements, no "10 prompts that will change your life". Just the things we write that we'd want to read.

One-click unsubscribe. We never share your address. Powered by our own infrastructure, not a tracker.

Other migration guides

Python frameworkFrom LangChain/migrate-from/langchainMulti-agent orchestrationFrom CrewAI/migrate-from/crewaiAutonomous goal pursuerFrom Auto-GPT/migrate-from/autogptHosted assistant APIFrom OpenAI Assistants API/migrate-from/openai-assistants