AI Integration

OpenToggl's practical AI integration path is not a vague "AI-ready" claim. It is a tool-driven workflow built around the maintained CorrectRoadH/toggl-cli project.

That repository provides two concrete surfaces for agents:

• a toggl CLI for day-to-day operations
• an installable SKILL.md package for agent environments

The real integration path

The toggl-cli fork explicitly positions itself as an actively maintained fork focused on AI agent friendliness and automation workflows. Its README also exposes a one-command skill install flow.

For agent environments, the documented install command is:

npx skills add CorrectRoadH/toggl-cli

For direct CLI usage, the documented npm install path is:

npm install -g @correctroadh/toggl-cli
toggl --help

What the CLI gives agents

The installed toggl binary gives agents a narrow, concrete command surface instead of forcing them to handcraft raw HTTP calls for every operation.

Based on the published skill and README, the CLI covers:

• Track API v9
• time entry operations like start, stop, continue, running, show, edit, and delete
• workspace resource operations like list, create, rename, and delete
• user/profile commands like me and preferences
• organization inspection commands like organization list and organization show

That is much more useful for agent workflows than a generic "AI API" promise, because the agent gets a tested, task-shaped interface for common operations.

OpenToggl-specific relevance

The same toggl-cli project also documents support for OpenToggl self-hosted.

Its interactive auth flow lets the user choose between:

• official Toggl Track
• OpenToggl self-hosted

And for OpenToggl it expects the full API URL, for example:

https://your-instance.com/api/v9

That means the intended agent flow is:

1. install the skill or CLI
2. authenticate once
3. point the tool at either official Toggl Track or an OpenToggl deployment
4. let the agent use stable CLI commands for normal tracking workflows

Why use CLI plus skill instead of only OpenAPI

Using toggl-cli plus SKILL.md gives agents:

• command shapes that already match common user tasks
• built-in auth flow
• local caching for read-heavy operations
• simpler automation loops than hand-assembling every request from scratch
• a reusable skill package that can be installed into agent environments

The README for toggl-cli specifically calls out local HTTP response caching and automatic cache invalidation when data changes. That matters for agent loops because it reduces repeated read calls while keeping mutated state fresh.

Where OpenToggl itself still matters

The CLI/skill is the execution layer, but OpenToggl still provides the product and deployment target:

• OpenToggl defines the API and product surface the tool can target
• OpenToggl self-hosting gives you an instance your agents can run against under your control
• OpenToggl docs and OpenAPI files explain the semantics behind those commands

Recommended setup for agent workflows

If you want an agent to work with OpenToggl, the practical setup is:

1. Deploy or run OpenToggl.
2. Install CorrectRoadH/toggl-cli as a CLI or skill.
3. Authenticate the tool against your instance.
4. Use the CLI command surface for normal operational tasks.
5. Fall back to OpenAPI and product docs when the agent needs lower-level contract detail.

References

• CorrectRoadH/toggl-cli repository
• skills/toggl-cli/SKILL.md
• OpenToggl product docs and OpenAPI contracts

OpenToggl does not need a separate AI product line if the agent path through CLI, skill, contracts, and self-hosted runtime is already solid.