Agents — Autonomous AI Participants
Filesystem-paired, autonomous agents — lifecycle management, rules, skills, templates. How PMs select, onboard, train, give feedback to, and performance-manage AI agents.
Stack position: Layer 4 (highest autonomy) — agents that operate independently with their own context, tools, and decision-making.
Sub-domains
System Design — Agent System Design
Patterns, techniques, and mental models for designing and configuring agent systems, organized by domain of practice.
Instruction Design — CLAUDE.md, agents.md & behavioral configuration
- Structured Context Loading — Purpose-built files loaded before each interaction to align agent behavior
- Knowledge Capture as Side Effect — Design agent systems so knowledge capture is a byproduct of corrections
solid
Skills — Discrete agent capabilities
- Meta-Skill Pattern — Build a “skill that builds skills” to bootstrap agent capabilities consistently
Supervision — Human governance of agent execution
- Stepwise Task Execution — One-task-at-a-time execution with pause-and-approve checkpoints
- Progressive Tool Disclosure — Revealing MCP tools in layers to combat choice paralysis and hallucination
- Agent-Mediated Self-Reflection — Using agents to observe behavioral patterns from digital exhaust
Architecture — Structural foundations
- Filesystem as Agent State — Agent architecture = filesystem (state) + LLM (orchestrator)
- Agent as Cross-Tool Workflow Hub — Local agent + MCP integrations as orchestration layer
Harnesses — Platform-Specific Knowledge
Knowledge and practices specific to individual agent platforms — setup, configuration, capabilities, best practices.
- Cursor: Structured AI Development — Three-file rule system, @-includes, MCP config, and Repo Prompt
Managing Agents — The Human-Agent Management Relationship
The practice of managing AI agents as autonomous participants — deciding what to delegate, how to delegate effectively, evaluating output, and developing agents over time. Distinct from system design (building agents) and harnesses (platforms) — this is about the relationship between human and agent.
Thesis: Talent-to-Direction Scarcity Shift — AI inverts traditional scarcity: talent is abundant, knowing what to ask for is the bottleneck
Task-Agent Fit — When to delegate:
- Delegation Decision Framework — Three-variable equation for when delegation pays off
Delegation Craft — How to delegate:
- Delegation Documentation as Agent Prompts — Every field’s delegation docs are AI prompts; common elements pattern
Evaluation & Feedback — Planned, no entries yet
Agent Selection & Onboarding — Planned, no entries yet