Ghost in the Machine: DEADLOOP, a Local-First Reliability Monitor for AI Agents

Description and use

DEADLOOP is a local-first reliability monitor for AI agents that detects repetitive tool calls and loops, issues evidence-backed incidents, and lets humans intervene without altering agent behavior.

Problem

AI agents often get trapped in repetitive tool calls, unproductive loops, or circular workflows, wasting tokens and time.

Outcome

DEADLOOP detects and reports repetitive agent behavior locally, enabling humans to intervene without external API calls, preserving data privacy and reducing token waste.

Workflow

• 1. Agent emits tool-call events to DEADLOOP via DeadLoopClient.send_event.
• 2. DEADLOOP ingests events, stores them in SQLite/PostgreSQL.
• 3. Detector logic evaluates traces for loop patterns and session budgets.
• 4. If a finding is detected, DEADLOOP generates an incident with event IDs, metrics, confidence, and uncertainty.
• 5. Human reviews incident, can acknowledge, dismiss, request pause, or approve pause.
• 6. If pause approved, agent execution is halted without claiming control.

Evidence

DEADLOOP ingests its tool-call trace and answers: Is this repetition, or legitimate retrying?

Source

The core is fully deterministic; an optional open-weight model only explains the findings. It never changes a finding or executes an intervention.

Source

It runs offline. Detection, risk scoring, ingestion, and incident management are deterministic and local. No API key is required for the core product to function.

Source

The sample simulator generates real execution events through the same ingestion, persistence, detection, and incident paths as production.

Source