Alternate model turns and action execution.
agent_turn_executionRECORD STATUSCandidate · v0.1.0The proposed contract drawn as participant roles, received calls, and numbered dependencies.
The same roles and calls as a class collaboration and a numbered message sequence.
Where the mechanism applies. All four dimensions belong to the proposed match, not just its vocabulary.
The obligation that clients, operators, and auditors should be able to rely on.
A one-shot model call with no action cycle does not satisfy the pointcut.
The hot spot is admitted only where the protected semantics survive.
Prompt, tool set, stopping predicate, and turn result encoding.
Test normal completion, immediate malformed output, and retry exhaustion.
Has the loop moved from ad hoc control to a typed runner or state machine?
An illustrative prompt and a small design sketch for the proposed contract. Names and APIs are invented; this is not a repository excerpt.
Work in a bounded loop. On each turn: call the model, execute any requested tool actions, record each observation, then ask the termination policy whether to continue. Return exactly one terminal result, or stop with a typed exhaustion error.def run(task, model, tools, policy):
history = [user_message(task)]
for turn in range(policy.max_turns):
reply = model.respond(history)
history.append(reply.as_message())
for action in reply.tool_calls:
observation = tools.execute(action, turn=turn)
history.append(observation.as_message())
decision = policy.decide(history, turn)
if decision.stop:
return decision.result
raise LoopExhausted(turns=policy.max_turns)Illustrative design sketch. The runner, not the prompt, owns turn identity and the terminal decision. This example is a teaching aid, not implementation evidence or a conformance claim.
The frozen rule that retrieves this record, the canonical source exemplar, and the control that should stay unmatched.
agent_turn_execution(?i)\bwhile\s+(?:true|not\s+done)|\bmax_(?:turns|steps|iterations)\b|\brun_(?:loop|until)\bFrozen expression, v0.1.0. A match is a retrieval cue, not a confirmed implementation.src/agents/run_internal/run_loop.py:987Support: tests/test_agent_runner.py:1Scope-anchored review. pohe-agent analyze <repo> --pattern P01 anchors the cue to its enclosing declaration and checks a separate operational signature for all four pointcut dimensions. Missing roles and negative-control risks remain visible; a structural match is not E2/E3. Calibration report ↗
Up to five public files retrieved by the frozen cue and pinned to the audited commit. A match is a place to inspect, not a confirmed implementation.
src/agents/run_internal/run_loop.py:987CANONICAL SOURCE02aaif-goose/goosecrates/goose/examples/agent.rs:61RETRIEVAL CUE03affaan-m/ECCskills/skill-comply/scripts/grader.py:94RETRIEVAL CUE04agno-agi/agnolibs/agno/agno/company/manager.py:328RETRIEVAL CUE05Aider-AI/aideraider/coders/base_coder.py:882RETRIEVAL CUEEach layer answers a different question. Do not substitute retrieval for review.
A concrete source anchor is available for this proposed record. It is one exemplar, not evidence that all repositories—or every contract field—conform.
Review boundary: this candidate and its specification come from an AI-assisted study. Independent human validation remains open.
97.2% of repository families in the frozen panel have at least one current cue, historical cue, or canonical source layer. This is not confirmed prevalence.
Current lexical-cue repositories: 39 / 4186bd29878b30src/agents/run_internal/run_loop.py:987tests/test_agent_runner.py:1Explicit agent turn/tool streaming loop.
(?i)\bwhile\s+(?:true|not\s+done)|\bmax_(?:turns|steps|iterations)\b|\brun_(?:loop|until)\bThis expression is a discovery instrument only. A match is a retrieval cue—not a confirmed implementation, pointcut match, or contract proof.