Fix readme curl examples

[muhsinking]

No description provided.

[runpod-Henrik]

QA Report

Status: PASS
PR: #222 — Fix readme curl examples
Agent: flash-qa (PR mode)

Scope

Docs-only PR — changes a single file: src/runpod_flash/cli/utils/skeleton_template/README.md. No source code, tests, or configs modified.

Diff Analysis

QB routes (gpu_worker, cpu_worker):

• Old: {"message": "Hello GPU!"} — missing both the input envelope and input_data wrapper
• New: {"input": {"input_data": {"message": "Hello GPU!"}}} — correct
• Verified: QB routes use make_wrapped_model which requires {"input": <inner>} envelope, and the function signature gpu_hello(input_data: dict) means the inner model has an input_data field. The fix matches the actual server behavior.

LB route (lb_worker/process):

• Old: {"input": "test"} — incorrect envelope; LB routes do NOT use the input wrapper
• New: {"input_data": {"message": "Hello from LB!"}} — correct
• Verified: LB routes use make_input_model directly (no make_wrapped_model), so the body maps to function params directly. For process(input_data: dict), the body should be {"input_data": {...}}.

Markdown syntax: No issues. All curl examples are syntactically valid shell commands with proper line continuations.

Test Sanity Check

Skipped — docs-only change to a skeleton template README does not affect any Python source or test files. CI will confirm independently.

Recommendation

MERGE — The curl examples are corrected to match the actual dev server request schemas. All three fixes (GPU QB, CPU QB, LB) are verified against the codegen in run.py and the helper functions in _run_server_helpers.py.

---

Generated by flash-qa agent