Attune Harness FAQ

Attune AI combines specification development, AI workflows, project memory, retrieval and verification tools. It is a separate project from Attune Harness, released under Apache 2.0. Provider access and setup depend on the workflow you choose.

Specify, Ground, Build, Remember and Verify describe a way to organize development: agree requirements, consult project sources, implement tasks, retain useful findings and check the result. This sequence is a working practice; it does not guarantee correct output or enforce every step in every workflow.

Attune AI brings together project memory, structured forms, AI workflows, retrieval through attune-rag and verification tools. Their availability and behavior depend on the installed version and the integration in use. Review the selected workflow and its checks before relying on its output.

The retained Attune AI documentation describes workflows, MCP tools, skills, wizards and template kinds. Use the reference for your installed version and inspect its available commands; a website count is not evidence that a capability is available in your environment.

attune-gui was the local dashboard that sat on top of the attune-ai framework. The project was parked in July 2026: the repository is archived and no further releases are planned. Released versions stay installable from PyPI (pip install attune-gui), and the previously announced fold into attune-ai[gui] was cancelled. See the archive notice for details.

A dashboard is not required for terminal workflows. CLI and plugin capabilities depend on your installed version and host. The earlier standalone attune-gui dashboard is archived; see the archive notice before relying on an existing installation.

Attune AI is open source under Apache 2.0. Hosted model calls and other external services may incur separate charges. Check the authentication route, provider terms and permissions for the workflow you intend to run.

Apache 2.0 permits use, modification and redistribution under its conditions. Redistribution obligations include preserving the license and relevant notices, and marking changed files. Read the license for the full terms, including patent provisions and warranty limits.

Yes, Apache 2.0 permits commercial use subject to its conditions. Preserve required license and attribution notices when distributing the software. Model access and other services have their own terms and charges.

attune-rag retrieves project sources to supply context and citations for review. Check the cited material and the resulting answer. A retrieval evaluation measures a particular corpus, query set and configuration; it does not establish accuracy for every generated answer.

Verification tools can check properties such as imports, command-line flags, links and capability counts against project sources. Inspect which checks ran, their results and any unchecked behavior. Passing those checks does not establish that all generated content is correct.

Local memory storage does not mean every Attune AI workflow stays on your machine. Hosted model calls can send the supplied code, prompts or task context to a provider. The location of Redis and embedding services depends on the configured setup. Review the workflow and its destinations before execution.

Authentication depends on the host and workflow. A host-native Claude Code integration can use its supported login; a direct API route needs the credentials and billing required by that provider. Check the selected route before execution rather than assuming a subscription covers every call.

Attune AI requires Python 3.10 or later. Check its installation guide and the requirements of your selected workflow, host and operating system. Python package availability does not establish compatibility with every IDE or integration.

Wizards are guided, multi-step workflows. Their questions, analysis and actions depend on the selected wizard. Consult the wizard reference for your installed Attune AI version and check its permissions before running it.

Use the wizard guide for your installed Attune AI version. Host commands and Python callbacks need the setup described by that guide; consult the retained documentation before running an example.

Yes! Two approaches: (1) YAML-based — create a .attune/wizards/my-wizard.yaml file with step definitions, no Python required. (2) Python-based — subclass BaseWizard for advanced logic like workflow delegation, conditional steps, and custom result processing. See the Custom Wizard Development guide.

Attune AI provides workflows for code review, security scanning, test generation, bug prediction and refactor planning, alongside specification and memory tools. Use them to support development, then inspect the output and run checks appropriate to the change.

The spec engine runs via /spec. It guides you Socratically through requirements, design, and tasks with an approval gate, so a feature starts from an agreed spec rather than an open-ended prompt — the first stage of the reliability loop.

The current Attune AI license is Apache 2.0. If you use an earlier release, inspect the license distributed with that release. This reference does not claim a reason for the historical licensing change.

Use the project documentation and GitHub community channels. Report Attune AI bugs in its issue tracker; use the Harness issue tracker for Harness. A pilot or support arrangement would need a separate discussion and agreement.

Report bugs via GitHub Issues. Include your environment details, steps to reproduce, and expected vs actual behavior.

Yes! We welcome contributions. Check out our GitHub repository for contribution guidelines. The framework is fully open source under Apache 2.0, making it easy to fork, modify, and contribute back.