Ignore local runtime artifacts and fix large tool result handling

[AminMahpour]

Summary

• Ignore .langgraph_api/, .pytest_cache/, and .playwright-mcp/ in git
• Stop tracking the generated .langgraph_api/* checkpoint and store files
• Route DeepAgents large tool results into the project-local artifact backend and materialize stale state-backed results for read_file

Testing

• uv run pytest -q passed
• Added regression coverage for artifact routing, workspace path mapping, and state-backed large tool result materialization

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 80ddaa147a

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Keep default agent bound to repo MCP server

Switching [agent].mcp_servers to "playwright" while the repo server block is commented out removes the default repo_* tool surface that this project’s documented flows rely on (for example /repo-readme and the repo-researcher path), so a fresh checkout now loses core repository tooling unless users manually reconfigure deepagent.toml. Keep repo enabled by default (or migrate all dependent flows/docs in the same change) to avoid a runtime regression in the default setup.

Useful? React with 👍 / 👎.