docs: add Cursor Cloud dev-environment setup notes to AGENTS.md

[Hmbown]

Summary

Sets up and verifies the CodeWhale development environment for Cursor Cloud agents, and records the non-obvious cloud-VM caveats in AGENTS.md under a new ## Cursor Cloud specific instructions section. No product code changed — only AGENTS.md.

Key learnings captured:

• libdbus-1-dev is a required system build dep (crates/secrets OS keyring).
• rustup default stable must be set so shells spawned outside the checkout (verifier background jobs, sub-agent worktrees) can find rustc.
• Two codewhale-tui subagent tests fail only because the checkout sits directly under / (unwritable parent; /workspace gets discovered as a nested repo when walking up). Not code bugs.
• How to run the agent without provider API keys via keyless vllm/ollama pointed at a local OpenAI-compatible endpoint.
• The codewhale dispatcher needs its sibling codewhale-tui co-located.

Environment verification

• Rust 1.97.0 stable (satisfies MSRV 1.88); installed libdbus-1-dev.
• Built the workspace (cargo build), both binaries report 0.8.68.
• Ran the agent end-to-end against a keyless local OpenAI-compatible mock LLM: plain one-shot codewhale exec, and the full agentic loop codewhale exec --auto (model → exec_shell tool call → shell execution → tool result → final streamed answer).

Testing

• cargo fmt --all -- --check
• cargo clippy --workspace --all-targets (clean)
• cargo test --workspace — 6384 passed; 3 failures are env-specific (repo at /workspace) / the documented run_verifiers_background_* flake, all explained in the new AGENTS.md section.

Checklist

• Updated docs or comments as needed
• Added or updated tests where relevant
• Verified TUI behavior manually if UI changes
• Harvested/co-authored credit uses a GitHub numeric noreply address

  

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[claude]

Claude finished @Hmbown's task in 55s —— View job

---

Review: docs-only change to AGENTS.md

• Gather context (PR is merged; single-file diff, AGENTS.md +34/-0)
• Review the diff
• Verify factual claims against the codebase
• Post findings

Summary: This is a documentation-only addition (a new ## Cursor Cloud specific instructions section in AGENTS.md). No product code, tests, or provider/model routing logic is touched, so categories 1–2 and 5 from the review checklist (correctness bugs, provider/model-prefix safety, security) don't apply here.

I spot-checked the specific technical claims made in the new section against the current tree, and they all hold up:

• libdbus-1-dev / OS keyring build dep — crates/secrets/Cargo.toml pulls in keyring with the linux-native-sync-persistent feature on Linux, which uses the secret-service/dbus stack, consistent with the stated dependency.
• DEEPSEEK_TUI_BIN override — confirmed in crates/cli/src/lib.rs (dispatcher's sibling-binary lookup honors this env var, with matching tests), so the fallback instructions are accurate.
• The two named env-specific test failures — git_repo_root_reports_attempted_paths_when_no_repo_found and create_isolated_worktree_reports_friendly_error_when_no_repo_found both exist in crates/tui/src/tools/subagent/tests.rs, matching the described /workspace-under-/ cause.
• run_verifiers_background_* flake — already documented elsewhere in AGENTS.md as a known pre-existing flake, consistent with how this PR references it.

Reuse/simplification (3): N/A — purely additive prose in an existing doc file, no duplicated logic.

Tests (4): N/A for a docs change; the PR's own testing checklist (cargo fmt, clippy, cargo test --workspace) is appropriately scoped and the reported 3 failures are explained as environment-specific rather than swept under the rug.

No issues found. This is a well-scoped, accurate, low-risk documentation PR that captures genuinely non-obvious cloud-VM setup caveats (build dep, rustup default, sibling-binary co-location, keyless local-provider testing) for future Cursor Cloud agents working in this repo.