Update README.md

feat: use mcp server to write test

chore: fix diagnostics

refactor(agent): unify LLM tool call handling with strong types and extraction

fix(tests): robustify qwen3-coder agent tests to pass with LLM JSON output quirks

docs(readme): document requirements and steps for running Ollama locally for integration tests (Qwen3-coder and Hello, world! test)

feat(agent): green TDD for agent actor with Ollama integration and deterministic 'Hello, world!' response. System prompt and user prompt engineered for strict output; full test suite passing.

refactor(git_client): remove deprecated commit_and_pr workflow and associated test

refactor(agentleman): separate git operations and PR creation using git_client and git_host actors

test(git_client): add PushBranchInput message/event, TDD unit test, and suppress git branch default hint

test(git_host): remove poll/wait/retry hacks—PR test now robust with correct commit, branch, and arg order

feat(agentleman): fully TDD-integrated happy flow orchestration with real PR, test isolation via UUID branch, robust error events, artifact cleanup, and assertion of PR existence via GitHub API. Test is robust and ids are dynamic. See git_client.rs for reusable cleanup.

chore: rm non-executable doc test

docs: document actor trait with executable example

refactor: update actor, developer, and issue_collector modules as part of ongoing improvements (see git_client refactor for details)