Track H: IDE and CLI

Track H introduces developer tooling: VS Code workflow preview, local CLI execution, and comparison between CLI trace output and SDK results.

Goal

Open a graph workflow in VS Code, view the graph, run a workflow locally via CLI, and compare CLI output to SDK output. First contact with ArcFlow developer tooling beyond raw scripts.

Prerequisites

Step 1: Open graph workflow in VS Code

Install or launch the ArcFlow extension from extensions/vscode-arcflow/. Open a workflow definition:

Use the extension graph panel to inspect nodes and edges. Confirm entry node and conditional edges match the Python DSL in reflection loop.

Step 2: Run workflow via SDK (baseline)

python examples/graph/reflection_loop.py

Record run_id and step_count from stdout.

Optional assertions from Track A pattern:

assert result.status == "completed"
assert result.run_id

Step 3: Run via CLI

From repository root, initialize a minimal project if needed:

cargo run -p arcflow-cli -- init my-track-h --language python

Run a workflow file through CLI (exact subcommand depends on your checkout; common pattern):

cargo run -p arcflow-cli -- run examples/graph/reflection_loop.py

If your CLI version expects a manifest path, point at the generated project or script per extensions/vscode-arcflow/README.md and CLI help:

cargo run -p arcflow-cli -- --help
cargo run -p arcflow-cli -- run --help

Pass criteria: CLI exits zero and prints run summary with run id.

Step 4: Compare CLI trace to SDK trace

Export trace for the run id from either path:

cargo run -p arcflow-cli -- trace YOUR_RUN_ID --format json --verbose

Compare event kinds to SDK:

kinds = {e.get("event_kind") for e in result.trace_events}
print(sorted(kinds))

Expect matching lifecycle and graph node kinds for the same logical workflow. Token counts and durations may differ slightly between invocations but structure should align.

Step 5: VS Code run integration (optional)

If the extension provides run commands, trigger run from the editor and open trace view. Compare displayed graph highlight order to GraphNodeStarted sequence in exported JSON.

Verification checklist

Expected output

SDK:

CLI trace export: JSON array of events with kind or event_kind fields per bridge version.

Troubleshooting

What you learned

Track H connects authoring (IDE graph), execution (CLI), and observability (trace export). Teams use this loop for local debugging before promoting workflows to server registry or static publish.

Next steps