Skip to content

Awesome list for AI agent harness engineering: tools, patterns, evals, memory, MCP, permissions, observability, and orchestration.

1.3 KB
Raw Blame

PLAN.md

Task planning artifact. Created at the start of a task; updated as milestones are reached. The agent should update this file throughout execution, not just at the end.

Task

Context

Approach

Milestones

Mark each milestone [x] when complete. Do not mark complete until the verification gate passes.

  • M1: — | verify: <command or check>
  • M2: — | verify: <command or check>
  • M3: — | verify: <command or check>
  • Final: all tests pass — verify: <test command>

Scope boundaries

In scope:

Out of scope (explicitly excluded from this task):

Open questions

Risks

Notes


Created: YYYY-MM-DD