Digitorn
Digitorn is an interpreter for AI agents. Install it once, write a
YAML manifest, run it. Same model as python for .py or node
for .js.
app:
app_id: hello
agents:
- id: main
role: assistant
brain:
provider: deepseek
model: deepseek-chat
backend: openai_compat
config:
api_key: "{{env.DEEPSEEK_API_KEY}}"
base_url: "https://api.deepseek.com/v1"
system_prompt: "Reply with one short sentence."
digitorn dev deploy hello.yaml --scope user
digitorn dev chat hello -m "ping"
# pong
That YAML is the entire app. No Python loop to wire up. No SDK glue code. No context-window bookkeeping. No tool dispatcher to implement. The interpreter handles all of it.
Why declarative
Every layer of the stack has moved from imperative to declarative because the declarative version scales better, audits cleaner, and removes glue code:
- Bash scripts → Terraform
docker run→ Kubernetes manifests- jQuery → React
- Manual SQL → ORM and migration specs
AI agents are the next layer. Writing the orchestration loop, the tool dispatcher, the retry policy, the sandbox, and the cost cap by hand is the bash-scripts-for-infra of the agent era. Digitorn makes the manifest the source of truth and runs everything from it.
What the interpreter handles
A single digitorn start brings every primitive a production
agent needs:
- 23 modules: filesystem, shell, web, http, RAG, vector, database, LSP, MCP, channels (webhook / cron / email / slack / discord / voice / ...), memory, workspace, widgets, ...
- 16 middleware: secret masking, content filter, RAG injection, retry, circuit breaker, dedup, audit, ...
- Multi-agent orchestration with sub-agent spawning
- OS-level sandbox (Landlock / seccomp / Job Objects)
- Credentials vault (OAuth2, API key, mTLS, ...)
- Real-time event stream over Socket.IO
- Behavior engine with declarative rules
Adding a new capability to your agent is one YAML key, not a
pip install plus integration code.
Install
Windows
irm https://digitorn.ai/install.ps1 | iex
macOS / Linux
curl -fsSL https://digitorn.ai/install.sh | sh
The installer fetches Python 3.12 via uv,
installs the digitorn CLI, registers a background service
(Windows Service / launchd / systemd), and starts the
interpreter on http://127.0.0.1:8000.
Already have Python 3.12?
pip install digitorn # or: uv tool install digitorn
digitorn service install
digitorn service start
First steps
digitorn doctor # check the environment
digitorn init my-app && cd my-app # scaffold a project
digitorn dev deploy app.yaml # push to the local interpreter
digitorn dev chat my-app # interactive chat
Full reference: docs.digitorn.ai.
Requirements
- Windows 10+, macOS 12+, or a recent Linux distro
- Python 3.12 (the installer fetches it via uv if missing)
- 2 GB free disk space for the model cache (embeddings, ONNX runtimes)
- Outbound HTTPS for LLM providers and MCP servers you choose to use
Operating the interpreter
digitorn service status # is it running?
digitorn service logs # last 50 lines
digitorn service stop # stop until next boot
digitorn service start # start again
digitorn service uninstall # remove the service
Logs:
- Windows: Event Viewer (
Applications and Services Logs > DigitornDaemon) - macOS:
~/Library/Logs/digitorn/ - Linux:
journalctl --user -u digitorn
The interpreter also writes structured logs to ~/.digitorn/logs/.
Development
git clone https://github.com/mbathe/digitorn-bridge
cd digitorn-bridge
uv sync # or: poetry install
uv run digitorn start # foreground, with the venv's deps
License
Apache 2.0. See LICENSE.
Metadata
Release files for digitorn 1.0.3
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| digitorn-1.0.3.tar.gz | 2.5 MB | With provenance | Details for digitorn-1.0.3.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| digitorn-1.0.3-py3-none-any.whl | Python 3 | none | any | Details for digitorn-1.0.3-py3-none-any.whl |
Total release size: 5.5 MB
Release files / digitorn-1.0.3.tar.gz
| Download URL | digitorn-1.0.3.tar.gz |
|---|---|
| Size | 2.5 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
4c60579d257e89a659da8e817f9aa03a81b37a459d318bb14213385275890818
|
|
BLAKE2b-256 checksum How to use checksums |
49cbcf0cbc8e493f62dfc9e01dcd40df82ae392cf7ec3650489a75a6d4c0b483
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 21, 2026.
Transparency logRelease files / digitorn-1.0.3-py3-none-any.whl
| Download URL | digitorn-1.0.3-py3-none-any.whl |
|---|---|
| Size | 3.0 MB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
6e93f966af3d3ebbc7610af3c717dc4b5b5c69717699de3915eb288371ea3138
|
|
BLAKE2b-256 checksum How to use checksums |
e6e54148d9ec3e820a5729d2a462dbfc3c62b65157f951d25ca4a8cad54ef391
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 21, 2026.
Transparency logSecurity for digitorn 1.0.3
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for digitorn-1.0.3.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 21, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.