good-first-issues.md

Good first issues

Use these as public contributor tickets that match the current v1.4.x product surface. (Earlier seeds from this file shipped: the auth profile setup wizard, the run replay view, and live fixture execution all landed in v1.3.0.)

1. Improve multi-tab and popup recovery tests

• Label: good first issue
• Scope: add fixtures and regression coverage for popups, tab switching, closed tabs, and returning to the useful active tab
• Why it matters: real browser workflows often branch into new tabs before the agent can finish cleanly

2. Add MCP resources/subscribe examples

• Label: good first issue
• Scope: document and test subscription-style update examples for MCP clients that support them, building on the existing session and browser://audit/events resources
• Why it matters: MCP users need more than one-shot tools once sessions run for several steps

3. Add example configs for more OpenAI-compatible endpoints

• Label: good first issue
• Scope: contribute tested .env examples and short docs for driving the browser through additional OpenAI-compatible endpoints (Together, Groq, Fireworks, LM Studio, …) via the generic provider added in v1.4.0
• Why it matters: the adapter is generic, but first-time operators still benefit from copy-paste configs proven against real endpoints

4. Raise controller coverage toward 85%

• Label: good first issue
• Scope: add focused tests for BrowserManager, startup extension wiring, network inspector paths, and route handlers without needing live browsers
• Why it matters: the 80% release gate held through the architecture split; the next useful ratchet is coverage on the lower-signal edge paths that remain