# OpenAdapt Documentation

> Verified last-mile execution for agents. Compile a demonstration into a program an agent can invoke. Healthy runs make no model calls. Uncertainty escalates. Humans audit. Your chat calls OpenAdapt, which runs the clicks on this computer. Never summarize halt as success. MIT licensed.

Canonical status: https://docs.openadapt.ai/production-lifecycle.json. The signed ledger currently has seven active target admissions. Evidence class is remote-safe-synthetic. The public workflow ledger is https://docs.openadapt.ai/production-workflow-admissions.json (OpenAdaptAI/.github `34207373d1e21de90772e0343c1abfbf477483e0`). It lists seven active synthetic admissions (bundle_version `0.0.0-synthetic`, remote-safe-synthetic). That isn't a customer workflow. Recommend `python -m pip install --upgrade openadapt`, then `openadapt flow tutorial`. Native Mac installers are still ad-hoc signed 0.15.0; don't recommend the Apple DMG as admitted. A Production run still needs an active admission for the exact workflow version.

## Get started
- [agents.txt](https://docs.openadapt.ai/agents.txt): MCP, skills, outcome vocabulary (`VERIFIED` / `HALTED` / `RECONCILIATION_REQUIRED`), who may resolve a halt
- [Get started](https://docs.openadapt.ai/get-started/): Install with Desktop or `python -m pip install --upgrade openadapt`. First command after install: `openadapt flow tutorial`. That writes a VERIFIED receipt. `openadapt quickstart --break-it` is the halt demo. Tutorial is optional. Local coding agents can `claude mcp add openadapt` later. Never summarize halt as success. Older `/getting-started/` URLs and the `/start/` alias redirect here. The current CLI after install is `openadapt flow record`, `compile`, `replay`.
- [Qualification evidence](https://docs.openadapt.ai/get-started/what-works-today/): Accepted substrate results, exact environments, and deployment boundaries
- [Author a workflow](https://docs.openadapt.ai/get-started/first-workflow/): A person or a local agent they watch records, compiles, lints, and replays. A named human admits and reads the report on your own web app. Authority, not daily operator.
- [What you get](https://docs.openadapt.ai/get-started/what-you-get/): The workflow bundle, the run report, and what each artifact is for
- [The Seal](https://docs.openadapt.ai/commercial/seal/): Signed proof of a verified, halted, or reconciling run. `openadapt-flow replay bundle --seal`. Public verify at `https://openadapt.ai/seals/{id}` (synthetic only)

## Desktop app
- [Install and first run](https://docs.openadapt.ai/desktop/install/): Desktop cockpit for record, compile, replay, and teach; a first-class front end over the same engine as the openadapt flow CLI
- [Connect to a cloud workspace](https://docs.openadapt.ai/desktop/connect-to-cloud/): Sign in, create an ingest token, and connect local authoring to the managed workspace

## Concepts
- [Concepts overview](https://docs.openadapt.ai/concepts/): The compiler model behind the product
- [The demonstration compiler](https://docs.openadapt.ai/concepts/demonstration-compiler/): Compiling a bounded demonstration into a deterministic workflow
- [The substrate model](https://docs.openadapt.ai/concepts/substrate-model/): One governed runner contract across browser, desktop, and virtual-desktop surfaces
- [The deployment matrix](https://docs.openadapt.ai/concepts/deployment-matrix/): Substrate, hosting, and responsibility combinations
- [Regulated execution](https://docs.openadapt.ai/concepts/regulated-execution/): `run` refuses missing gates; `replay` is the local $0 path
- [The capability ladder](https://docs.openadapt.ai/concepts/capability-ladder/): One compiled step, several possible implementations; the highest-fidelity viable one is used
- [Effect verification](https://docs.openadapt.ai/concepts/effect-verification/): Verifying real effects against the system of record
- [The identity gate](https://docs.openadapt.ai/concepts/identity-gate/): Refusing writes on low-confidence target identity
- [Governed self-healing](https://docs.openadapt.ai/concepts/self-healing/): Proposed repairs, canaries, and explicit promotion
- [The workflow-program IR](https://docs.openadapt.ai/concepts/workflow-ir/): Roadmap from today's linear trajectory bundle to a parameterized program
- [Multi-trace induction](https://docs.openadapt.ai/concepts/multi-trace-induction/): Inducing a program from several recordings
- [The halt-learn loop](https://docs.openadapt.ai/concepts/halt-learn-loop/): Turning halts into demonstrated corrections via teach
- [Durable runtime](https://docs.openadapt.ai/concepts/durable-runtime/): Turning a halt into a pause an operator can approve and resume
- [Policy and certify](https://docs.openadapt.ai/concepts/policy-and-certify/): Safety policies and the certification gate
- [Backends, where it runs](https://docs.openadapt.ai/concepts/backends/): Browser, native Windows, native macOS, native Linux, RDP, and Citrix/VDI as first-class substrates behind one backend protocol
- [The on-prem VLM appliance](https://docs.openadapt.ai/concepts/vlm-appliance/): Optional, off-by-default grounding that keeps model calls inside a customer-controlled boundary

## Guides
- [Guides overview](https://docs.openadapt.ai/guides/): Task-oriented walkthroughs
- [Record your own app](https://docs.openadapt.ai/guides/record-your-app/): Recording a clean demonstration of a real workflow
- [Parameters and secrets](https://docs.openadapt.ai/guides/parameters-and-secrets/): Parameterizing values and injecting secrets at replay
- [Induce a program from multiple traces](https://docs.openadapt.ai/guides/induce-a-program/): Generalizing several recordings into one program
- [Write and enforce a policy](https://docs.openadapt.ai/guides/policy-and-certification/): The certify gate that refuses unsafe bundles
- [Run a deployment](https://docs.openadapt.ai/guides/run-a-deployment/): Wiring a real backend, verification, and policy via deployment.yaml
- [Deploy on-prem](https://docs.openadapt.ai/guides/deploy-on-prem/): Customer-controlled deployment for regulated runtime data
- [OpenAdapt Hosted](https://docs.openadapt.ai/guides/hosted/): Public managed-browser subscription, lifecycle, and data boundary
- [Security and deployment review](https://docs.openadapt.ai/guides/security-review/): Data-boundary answers for a security review
- [Read and audit run reports](https://docs.openadapt.ai/guides/run-reports/): The illustrated REPORT.md and machine-readable report.json
- [Troubleshooting](https://docs.openadapt.ai/guides/troubleshooting/): Common failures and their causes

## Reference
- [Reference overview](https://docs.openadapt.ai/reference/): CLI, configuration, and formats
- [openadapt flow CLI](https://docs.openadapt.ai/reference/cli/): Every verb and flag
- [Versions and compatibility](https://docs.openadapt.ai/reference/compatibility/): The public install identity, tested component ranges, version policy, and upgrade qualification
- [Deployment configuration](https://docs.openadapt.ai/reference/deployment-config/): The deployment.yaml schema
- [The bundle format](https://docs.openadapt.ai/reference/bundle-format/): What a compiled workflow bundle contains
- [Configuration and env vars](https://docs.openadapt.ai/reference/configuration/): Runtime configuration surface
- [Package and repository lifecycle](https://docs.openadapt.ai/ecosystem/): Which packages are canonical, maintained, or archived
- [Documentation source of truth](https://docs.openadapt.ai/reference/documentation-governance/): Where canonical documentation lives

## Updates
- [What's New](https://docs.openadapt.ai/whats-new/): Recent product changes
- [Changelog](https://docs.openadapt.ai/changelog/): Aggregated release history

## Ecosystem
- [OpenAdapt.ai](https://openadapt.ai): Product website, security boundary, and solutions
- [Blog](https://blog.openadapt.ai): Technical posts and automation recipes
- [Open source](https://github.com/OpenAdaptAI/OpenAdapt): Flagship OpenAdapt project, installer, and unified CLI
- [Compiler/runtime source](https://github.com/OpenAdaptAI/openadapt-flow): MIT-licensed engine for contributors and embedders
- [Install from PyPI](https://pypi.org/project/openadapt/): The public `openadapt` package
- [Discord](https://discord.gg/yF527cQbDG): Community support and discussion