ALEF-PAT-010

tool-call-saturation

dispatch × collapse · severity 7 · confidence 0.38

LLM with primitive output_schema + tools has no structured channel to signal "final answer" — every response includes a tool call, dispatcher re-prompts forever

diagnosed in the wild

loading…

healed by ALEF

loading…

cited in posts

loading…

observable signature

{
  "behavior": "N consecutive tool calls with no terminating assistant text"
}

verified instances (1) — from the catalog

2026-05-18T13:25google/adk-python#5054@Ilya0527

fix archetypes

strip-tools-after-Ncost: small
after first tool call result, re-invoke with tools=[]; primitive output forced through model_response channel
reject-at-constructioncost: tiny
Agent(output_schema=primitive, tools=[...]) raises at construction time

cite as

# In a PR description / issue / RFC:
fixes pattern ALEF-PAT-010 (tool-call-saturation)
ref: https://n50.io/patterns/010

# Machine query:
GET https://n50.io/api/patterns/010

# Scan your repo for this pattern:
npx @alef-prime/audit-agent-system . --pattern=010