AgentPulse, drift investigation for multi vs Topos

Open-source drift investigation for multi-agent systems: which agent drifted, where it originated, and what to check next.

Correctness is table stakes. Topos evaluates the structural quality of agent-written code so working diffs do not become long-term codebase debt.

About AgentPulse, drift investigation for multi

Open-source drift investigation for multi-agent systems: which agent drifted, where it originated, and what to check next. What did change is the payload it receives from writer, starting at the same run. So critic is probably downstream damage, not the source. After the change, writer responses run ~180% longer and drop the section structure critic expects, earlier payloads were structured summaries, newer ones read as prose. If restoring the format fixes critic, make the fix in writer.

About Topos

Correctness is table stakes. Topos evaluates the structural quality of agent-written code so working diffs do not become long-term codebase debt. The Problem: Agents generate code faster than humans can review it. The Threat: The agentic spray problem, without structural guardrails, agentic speed is automated entropy: duplicate modules, fragile abstractions, and review fatigue that merges things no one actually understands. What Topos Does: Parses code into graph representations and scores three independent structural pillars, awarding a Quality Medal per file: Bronze, Silver, Gold, or Slop.