long-runner-harness

Automate browser testing for long-running agents by bootstrapping a structured harness that orchestrates workflows, validates features in isolated sessions, and verifies UI correctness using Playwright MCP.

Session isolation in autonomous agent workflows prevents context bleed by executing each feature in a separate, secure context, ensuring deterministic dispatch and accurate browser-based validation outcomes.

Prevent false success reports in automated browser testing by enforcing external validation through real browser verification via Playwright MCP and collecting traces to cross-check autonomous agent execution outcomes.

Yes, you can use Playwright MCP to verify UI correctness by running each agent feature in an isolated session, validating visual fidelity and capturing execution traces for cross-session learning.

Orchestrate dual-agent coordination with dependency resolution by using an automated deterministic dispatcher that drives feature progression, manages rollback-safe commits, and maintains secure session isolation throughout.

Yes, you need bash, python3, git, and curl installed to execute the scripts that bootstrap the environment and run the automated long-running agent browser testing harness.