From first prompt to finished workflow.

Every run follows the same path. You define the work, Ottyr resolves the context, your chosen AI does the thinking, and Ottyr keeps the result.

Runs can begin with a click, a file drop, pasted clipboard text, a folder event, or another prompt. For multi-step work, chain prompts so each result becomes the next prompt’s input.

The main window organizes setup and review. Two always-on surfaces—the Command Bar and Prompt Sidebar—make your prompts available from anywhere on your Mac.

A prompt is a saved job: a clear set of instructions plus the profile, files, and output behavior needed to run it consistently.

Five ways to run

Projects group related prompts—one per client, product, or workflow is a useful starting point. Create projects from the sidebar; the Dashboard keeps the cross-project view.

Reports/{{VAR:project-name}}/{{VAR:date}}.md

An AI profile tells Ottyr how to execute a prompt. Keeping provider setup separate from prompt content means the same prompt can run through different tools.

A profile stores the provider, model, optional system prompt, and provider-specific settings such as executable path, working directory, API URL, API key, sessions, and extra flags.

Variables fill in values when a prompt runs. Use them in prompt text and output paths with {{VAR:variable-name}}.

Names are trimmed, case-insensitive, and normalized so spaces become hyphens. {{VAR:Project Name}}, {{VAR:project name}}, and {{VAR:project-name}} all resolve to the same value.

Write a weekly update for {{VAR:project-name}} on {{VAR:date}}.
Focus area: {{VAR:focus-area}}

Automatic variable reference

Input-file variables require an input file. Output-file variables require an output file path.

Snippets are reusable blocks for shared instructions, output formats, examples, policies, and project context. Define one under Snippets, then insert it from the prompt editor or type its reference.

{{SNIPPET:identifier}}

These four template commands cover dynamic values, reusable text, and multi-step workflows.

{{VAR:...}} and {{SNIPPET:...}} work in prompt text and output paths. Ottyr removes {{PROMPT:...}} and {{CHAIN:...}} from the model input and uses them to orchestrate the workflow.

Chaining lets a successful prompt start other prompts automatically. The full output of the parent run becomes input for the next prompt.

{{PROMPT:draft-follow-up}}

{{CHAIN:extract-tasks|prioritize-tasks|draft-email}}

Prompt codes are unique identifiers generated from the title and editable on the prompt screen. They may contain lowercase letters, numbers, dashes, and underscores. Multiple {{PROMPT:...}} commands can start several prompts from the same parent output; use {{CHAIN:...}} when each step should feed the next.

Good fits: draft then review; extract then summarize; classify then route; generate an artifact then create a short human-readable report. Configure chaining under Settings → Execution.

A watched-folder trigger runs one or more prompts when a matching file is added, updated, or deleted.

Added and updated files become input assets. For a deletion, the prompt receives text describing the deleted path and event.

Filter precisely

File groups include images, documents, PDFs, spreadsheets, presentations, audio, video, archives, and code.

The Command Bar is a Spotlight-style panel for finding and running prompts even when Ottyr is not the active app.

Output and errors appear in the panel. A common flow: copy an email, open the Command Bar, find “draft a reply,” and run it without leaving the app you’re in. The global shortcut is configurable in Settings.

The Prompt Sidebar is a floating dock pinned to a screen edge for fast, hands-on runs. Toggle it with ⌃⌥Space, from the menu bar, or keep it open in Settings.

Choose the dock edge in Settings or from the menu bar. The sidebar shows pinned prompts by default and can also search your library. Run now starts a prompt with no input.

Ottyr lives in the macOS menu bar, so core actions stay one click away.

Run History is the audit trail for manual, triggered, and chained work. It lets you revisit what ran, with which context, and what came back.

Artifacts captured alongside output

Search earlier outputs or failures, jump from a triggered run to its trigger, inspect chain lineage, export CSV from Settings, or purge runs older than the retention period. Artifact cards offer Open and Folder actions; prompt details show and can clear that prompt’s artifact folder.

Settings group app behavior, execution limits, history, exports, and global shortcuts into focused tabs.

Relative output names use the default output directory when a run has no input file to save beside. Old runs remain until you purge them using the configured retention period.

Both global shortcuts are configurable under Settings → Shortcuts.