Drift Signals

Built-in signals.

Zero assertions to write.

Spooled detects behavioral changes automatically from execution structure. No test code to maintain. Content-blind signals work without reading your prompts, responses, or customer data.

new_behavior_patterninfo

Cold start: fingerprint doesn't match any known intent. The system is learning a new execution path.

new_side_effectsmedium

Unauthorized tools detected — tools appearing in the current run that weren't in the baseline.

latency_spikeshigh

Average or max latency exceeds baseline bounds. Triggers on >50% increase or when max exceeds 2× the p95 envelope.

retry_explosionshigh

Excessive consecutive retries of the same tool after errors. Detects genuine retry loops, not pagination or batch patterns.

error_increaseshigh

Error rate increased significantly versus baseline. Catches agents that silently start failing more often.

tool_usage_changesmedium

Tool call count changed by more than 50%. Detects over-calling or under-calling of expected tools.

token_usage_spikehigh

Token consumption increased by more than 50%. Catches prompt bloat, unnecessary context, or model output changes.

component_latency_driftmedium

Latency increased for a specific component — llm:gpt-4, tool:search, http:api.example.com. Pinpoints the source.

tool_overusemedium

A tool is being called more than necessary. Detects redundant or circular tool invocations.

retrieval_regressionhigh

RAG retrieval quality degraded. Monitors retrieve, vector_search, search, and query tools for precision changes.

output_schema_drifthighPro

Output field schema changed — fields added or removed from tool responses. Catches breaking API contract changes.

Stop shipping blind.

$pip install spooled-ai
© 2026 GMH Labs LLC. All rights reserved.