Document async execManagedNe batch option for oclaw job polling.

Clarify large-batch background behavior and add async flag to the tool schema.

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
oliver 2026-08-12 22:17:16 +08:00
parent 0eb579609b
commit 23ca7bb980

View file

@ -653,7 +653,9 @@ HTTP_MCP_TOOLS: list[dict[str, Any]] = [
"(2) different CLI per NE (vendor/role) → ONE targets=["
"{ume_ne_id|ne_id, commands:[…]}, …] — do NOT fall back to one-NE loops. "
"Do NOT loop one-NE execManagedNe for multi-NE work. "
"Default read_timeout_sec=60; on timeout raise to 90–120 — do not blind-retry."
"Default read_timeout_sec=60; on timeout raise to 90–120 — do not blind-retry. "
"Large batches (≈4+ NEs) may auto-run async in oclaw: returns job_id immediately; "
"poll get_ne_exec_job. Pass async=true to force background, async=false to force sync."
),
"inputSchema": {
"type": "object",
@ -720,6 +722,13 @@ HTTP_MCP_TOOLS: list[dict[str, Any]] = [
"default": 4,
"description": "Parallel NEs for batch mode (ignored for single-NE).",
},
"async": {
"type": "boolean",
"description": (
"oclaw-only: true=background job_id + get_ne_exec_job; "
"false=force sync; omit=auto for large batches (~4+ NEs)."
),
},
},
"required": [],
"additionalProperties": False,