XIII · Cognition & IntrospectionEmerging★

Typed Tool-Loop Failure Detector

also known as Dispatch-Boundary Veto, Five-Mode Loop Guard, Tool-Call Pattern Detector

Lift tool-loop detection from prompt-level rules to a mechanical dispatch-boundary veto with typed failure modes and per-tool caps that returns a formatted refusal the model must consume.

This pattern helps complete certain larger patterns —

  • specialisesCircuit Breaker★★— Stop calling a failing dependency for a cooldown period after error rates exceed a threshold.

Context

A team is running an agent with a rich tool palette in which loop bugs — the agent calling the same tool over and over, or cycling through a small subset of tools without progress — can eat substantial budget before any safety net trips. Prompt-level instructions telling the model 'do not call X more than three times' are not actually enforced: the model can simply ignore them. A single global circuit-breaker on total tool calls catches the most extreme cases but hides the specific shape of the failure when it does fire.

Problem

Tool-explosion is named elsewhere in the catalogue as an anti-pattern, but naming it provides no mechanism to catch it. A single global circuit-breaker misses the shape of the underlying failure: a thirty-call canvas-action burst looks identical to thirty healthy file reads under a flat global counter, so the breaker either trips too often on legitimate bursts or too late on real failures. Prompt-level rules are advisory only, so the model can ignore them when it is most stuck. The team needs detection lifted from the prompt to a mechanical check at the dispatch boundary, with typed failure modes and per-tool caps that emit a refusal the model is forced to consume rather than silently retry.

Forces

  • Per-tool caps are noisy without good defaults.
  • A typed refusal must be formatted so the model can consume it as input rather than silently retry.
  • Global breaker is the backstop but should be the last to fire.
  • Detection windows must be tunable; too short trips legit work, too long drains money before tripping.

Example

A long-running personal agent has a canvas-action tool that occasionally enters a thirty-call burst when an interaction goes wrong. The global step-budget catches it eventually but only after thousands of tokens. The team adds a Typed Tool-Loop Failure Detector with per-tool caps: canvas-action is capped at four calls in a sixty-second window. When the burst starts, the fifth call returns a typed refusal `{mode: 'generic_repeat', observed: {...}}`. The model sees the refusal in its next observation and shifts to a different approach instead of pounding the same tool.

Diagram

Solution

Therefore:

A dispatcher pre-check function. On each tool call, append `(timestamp, tool_name, hash(args))` to a bounded rolling window. Evaluate five rules: (1) generic-repeat: same `(tool, arg-hash)` at least N times in window; (2) unknown-tool-repeat: call to unregistered tool at least M times; (3) poll-no-progress: same tool with no state change at least K times; (4) ping-pong: alternating between two tools at least J cycles; (5) global-circuit-breaker: total tool calls in window at least G. Each rule has per-tool overrides (for example a known-bursty tool capped lower than the default). On trip, the dispatcher returns `{error: 'tool_loop_detected', mode: <id>, observed: <stats>}` as the tool result. The model sees this in its next turn and must adjust.

What this pattern forbids. No tool call may bypass the dispatch-boundary loop check; a tripped detector blocks that specific call and returns a typed refusal that becomes the next observation, and the per-tool cap cannot be raised mid-session by the model.

And the patterns that stand alongside it, or against it —

  • complementsStep Budget★★— Cap the number of tool calls or loop iterations the agent is allowed within a single request.
  • complementsPre-Generative Loop Gate·— Before the next generation fires, detect divergence signatures (narration loops, frustration paths, repetition pressure) and inject a diagnostic steering hint into the prompt rather than veto the call.

Neighbourhood

Click any neighbour to follow the language. Scroll to zoom, drag to pan.

Used in frameworks

References

Provenance