[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"readme:bemyagent":3},"\u003Ch1>BEMYAGENT.md\u003C\u002Fh1>\n\u003Cblockquote>\n\u003Cp>\u003Cstrong>Mission\u003C\u002Fstrong>: Save tokens for the machine. Save orientation for the human.\u003C\u002Fp>\n\u003C\u002Fblockquote>\n\u003Cp>📖 \u003Cstrong>Website\u003C\u002Fstrong>: \u003Ca href=\"https:\u002F\u002Fbemyagent.md\" rel=\"nofollow ugc noopener\">bemyagent.md\u003C\u002Fa>\u003C\u002Fp>\n\u003Cp>BEMYAGENT.md is a lightweight, self-bootstrapping protocol that bridges the gap between humans and AI agents. Instead of forcing alignment through code reviews or rigid procedures, it creates a shared workspace where the machine thinks in structured files and the human validates at the right level of abstraction.\u003C\u002Fp>\n\u003Ch2>The Problem\u003C\u002Fh2>\n\u003Cp>When working with AI agents on complex projects, three things break down:\u003C\u002Fp>\n\u003Col>\n\u003Cli>\u003Cstrong>Context bloat\u003C\u002Fstrong> — The agent reads thousands of irrelevant lines, inflating costs and slowing down.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Silent drift\u003C\u002Fstrong> — The agent executes a task but drifts from the original intent. Nobody catches it until it's too late.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Validation fatigue\u003C\u002Fstrong> — The human must review every line of output because there's no structured checkpoint between \"done\" and \"delivered\".\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Ch2>The Solution: TTEV Workflow\u003C\u002Fh2>\n\u003Cp>BEMYAGENT.md provides a single markdown file (\u003Ccode>BEMYAGENT.md\u003C\u002Fcode>) that acts as a bootstrap prompt. When fed to an AI assistant, it generates a structured \u003Ccode>.bemyagent\u002F\u003C\u002Fcode> workspace:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>\u003Ccode>.bemyagent\u002Fdocs\u002F\u003C\u002Fcode>\u003C\u002Fstrong> — Permanent project memory (architecture, code map, tech stack, decisions).\u003C\u002Fli>\n\u003Cli>\u003Cstrong>\u003Ccode>.bemyagent\u002Fwork\u002F\u003C\u002Fcode>\u003C\u002Fstrong> — Tactical, volatile memory organized as a Hierarchical Task Network (HTN).\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3>Core Concepts\u003C\u002Fh3>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Concept\u003C\u002Fth>\n\u003Cth>What it does\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>\u003Cstrong>TTEV Workflow\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>Think → Task → Execute → Verify. A four-phase cycle where the agent strategizes, plans atomic steps, executes, and self-validates before notifying the human.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Lazy Loading\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>The agent never reads specs, drafts, or decisions during context restoration unless the current task explicitly requires them. Saves tokens by default.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Fractal Decomposition (HTN)\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>If a task is too large, the agent decomposes it into sub-tasks (e.g., \u003Ccode>work\u002F1\u002F1.1\u002F\u003C\u002Fcode>, \u003Ccode>work\u002F1\u002F1.2\u002F\u003C\u002Fcode>). Each leaf node gets its own TTEV cycle.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Context Saturation Check\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>Before executing, the agent verifies it has enough context (target files, expected behavior, constraints, dependencies). If too much is unclear, it asks instead of guessing.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Contextual DNA Mapping (CDM)\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>During planning, the agent embeds validation criteria directly into each task — scaled by complexity. Simple tasks get none; complex tasks get Drift sensors, Validation criteria, and Pivot triggers.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Symbiotic Validation\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>After execution, the agent evaluates its own output against the CDM criteria and produces a verdict (PASS \u002F PASS_WITH_CAVEATS \u002F FAIL) before presenting results. The human validates the \u003Cem>sense\u003C\u002Fem>, the agent has already validated the \u003Cem>form\u003C\u002Fem>.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Self-Registration\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>The agent configures the project's native rule files (\u003Ccode>.cursorrules\u003C\u002Fcode>, \u003Ccode>AGENTS.md\u003C\u002Fcode>, etc.) to read \u003Ccode>00-ai-rules.md\u003C\u002Fcode> at every session start.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\u003C\u002Ftable>\n\u003Ch3>Pacing Modes\u003C\u002Fh3>\n\u003Cp>The human controls how much autonomy the agent has:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>SEAMLESS\u003C\u002Fstrong> — The agent runs TTEV automatically. It only stops if verification finds issues.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>INTERACTIVE\u003C\u002Fstrong> — The agent pauses after THINK (plan approval) and after VERIFY (result approval). Two human gates.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>AUTO-CLI\u003C\u002Fstrong> — The agent switches AI models per phase (e.g., large model for THINK, fast model for EXECUTE).\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2>Usage\u003C\u002Fh2>\n\u003Col>\n\u003Cli>Drop \u003Ccode>BEMYAGENT.md\u003C\u002Fcode> into the root of your project.\u003C\u002Fli>\n\u003Cli>Ask your AI assistant to read the file and execute its instructions.\u003C\u002Fli>\n\u003Cli>The AI generates the \u003Ccode>.bemyagent\u002F\u003C\u002Fcode> directory structure and templates.\u003C\u002Fli>\n\u003Cli>Delete \u003Ccode>BEMYAGENT.md\u003C\u002Fcode> and start a fresh chat session (the bootstrap context is no longer needed).\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Cp>That's it. From this point on, the agent reads \u003Ccode>.bemyagent\u002Fdocs\u002F00-ai-rules.md\u003C\u002Fcode> at the start of every session and knows how to operate.\u003C\u002Fp>\n\u003Ch2>How It Works (The Files)\u003C\u002Fh2>\n\u003Cpre>\u003Ccode>.bemyagent\u002F\n├── docs\u002F                          # Permanent project memory\n│   ├── 00-ai-rules.md             # The protocol itself (agent reads this first)\n│   ├── 01-overview.md             # What the project does, quick start\n│   ├── 02-architecture.md         # System diagram, component roles\n│   ├── 03\n\u003C\u002Fcode>\u003C\u002Fpre>\n",1784240406344]