Hoist llm.ask prompt binding to avoid elevated `p`; prefer reasoning_content when content is empty; point live/tests at in-tree MARQDO_EXT; add USTC live gates and mock agent.step selection+resume evidence; view document --out for read-only browser embed checks. Design stays Partial (not Verified). Co-authored-by: Cursor <cursoragent@cursor.com>
1.1 KiB
1.1 KiB
markup, title, description, import
| markup | title | description | import | ||||||
|---|---|---|---|---|---|---|---|---|---|
| occlusion | workbench M01 live selection (USTC-compatible) | Explicit ModelSelection via llm.create; no llm.llm. Credentials from process env only. |
|
main
base = > sys.env_get name="OPENAI_BASE_URL" key = > sys.env_get name="OPENAI_API_KEY" mdl = > sys.env_get name="OPENAI_MODEL"
- not
base
print text=missing-OPENAI_BASE_URL sys.exit code=2
_ = 1
- not
key
print text=missing-OPENAI_API_KEY sys.exit code=2
_ = 1
- not
mdlmdl = "qwen3.8-chat" -
_ = 1
model = > llm.create model=mdl base_url=base api_key=key backend="openai-compatible" name=workbench-live
q = "Reply with exactly one English word pong"
reply = > model.ask prompt=q
text = > json.get value=reply key="text"
usage = > json.get value=reply key="usage"
pt = > json.get value=usage key="prompt_tokens"
print text=
text
pt
print text=usage-present
print text=usage-unknown