AI Assistant
Bring your own AI provider for PR summaries, the sprint assistant, and smart search in the command palette.
Providers
Open Settings → AI, enable AI features, pick a provider, paste your API key and optionally set a model (leave it empty for the provider default). Keys are stored in the macOS Keychain.
Claude
Create an API key at console.anthropic.com → API Keys.
OpenAI
Create an API key at platform.openai.com → API Keys. Set a Base URL for compatible APIs.
Langdock
OpenAI-compatible: enter your Langdock API key and the Base URL of your deployment.
Ollama Cloud
Create a key at ollama.com/settings/keys. Ollama Cloud is plan-based, so usage shows tokens but no per-token cost.
Kimi
Create a key at platform.moonshot.ai → API Keys. Use the Base URL field for api.moonshot.cn accounts.
You can also choose whether AI answers are written in English or in the app language.
Usage and cost history
Settings → AI → Usageshows today's calls, tokens and estimated cost. Usage history breaks usage down per day, per provider and per model, with cost estimates (providers with flat plans, such as Ollama Cloud, show tokens only).
Daily limits
Cap how much DevBar can use your AI account each day. When a limit is reached, AI features pause until the next day.
| Daily call limit | Maximum AI calls per day. 0 uses the default of 50. |
| Daily token limit | Maximum DevBar AI tokens per day. Empty means no limit. |
| Daily cost limit (USD) | Maximum estimated DevBar AI spend per day. Empty means no limit. |
Import Claude Code usage
DevBar can include your Claude Code token usage in the usage history by reading the local session logs. Nothing leaves your Mac.
- In Settings → AI → Claude Code, turn on Import Claude Code usage.
- Click Connect Claude Code logs… and select your
~/.claudefolder (or itsprojectsfolder) to grant access. - Use Scan now to refresh; the last scan time is shown next to the folder.
If macOS access expires the status changes to "Access expired" — reconnect the folder to resume importing.