agent-behavior

Standards for defining and evaluating agent behavior

⚡ High-frequency (~60Hz) in-browser behavior tree execution engine for autonomous AI agents. Reactive priority interrupts, safety guardrails & deterministic action replay over MCP.

The meta layer behind AI skills. Persistent behavioral overlays that make expected agent conduct explicit across changing tasks and capabilities.

Studying the gap between what agents know and when they act on it.

Give your coding agent a short break after real work, then measure what happens next. Universal agent skill & experiment harness.

Behavior-management plugin for DeepSeek Harness: tool-call discipline prompt section, failure-triggered parallelism convergence (pool drops to 1, auto-restores), consecutive-failure user intervention. Complements dsh-token-optimizer; works standalone.

The senior dev, unbundled — ten opinionated skills that make an AI coding agent push back, verify, attack its own code, and finish like a senior engineer.

High-performance routing engine that selects the best agent skill for a task and emits structured handoff decisions.

Claude Code 插件 | Cursor 插件 | Trae 插件 | Codex 插件 — AI 智能体行为准则：请示报告、督促检查、信息服务。跨 Claude Code / Cursor / Codex / Trae 的 agent 工作纪律插件。

text systems, behavioral contracts, AI protocols, and small tools for keeping AI-assisted work operational. I built these tools to reduce drift, force useful decisions, and keep messy AI-assisted work usable.

RL-style eval measuring intent/action divergence in frontier agents: model acknowledges a correction, then acts on the stale value anyway. 3 scenarios, 655 trials on claude-haiku-4-5, Sonnet 4.6, GPT-5.4, and Gemini 3.1 Pro Preview.

Agent authority guardrails for Claude Code & Grok — stop yield-back; drive reversible work yourself.

Toy 7. An elimination-filter landscape applying two structural constraints simultaneously to map which objective classes can persist under sustained optimization pressure — and which cannot. Includes a four-stage scenario engine and open-question frontier. Companion simulation for The Shape of What Does Not End — Series 2, Part 4.

Canonical home for AI Behavior Science research and the Founding Territory Paper

Lint AgentSkill folders before install, sync or release: YAML frontmatter, name vs folder, description length, dev-tree paths, model-run telemetry, README standard. CLI, /skill-lint command and GitHub Action.

Documentation-only, sanitised academic case study for a Unity maze-chase game AI project.

Toy 5. An interactive proxy decay simulator showing how optimization pressure erodes the modeling capacity required to distinguish proxy from territory — producing self-reinforcing V(t) degradation that becomes progressively harder to correct. Companion simulation for The Depth Constraint — Series 2, Part 2.

Composable behavior mutators for agent prompts - priority, routing, batching, topology

Toy 4. An AI-powered goal-chain classifier and invariance stress tester. Enter any goal, trace the instrumental chain to its terminal type, then probe whether the classification survives symbolic, blind, independent, and adversarial re-examination. Companion simulation for The Invariant Drive — Series 2, Part 1 of the Architecture of Thriving.

IAB — The First Workshop on Interpreting Agent Behavior @ NeurIPS 2026. Human-centered interpretation for understanding agents, humans, and interaction.