com um clique
plan
// Entry point for the three-tier planning system. Use when starting planning work at any level - project, milestone, or task.
// Entry point for the three-tier planning system. Use when starting planning work at any level - project, milestone, or task.
Add a new AI coding agent to Agent Sandbox. Creates all required files (Dockerfile, templates, CI, docs) and wires the agent into the CLI, proxy, and build system.
Break a milestone into discrete tasks. Use when starting work on a milestone that has been defined in the project plan.
Create a project plan for a new initiative. Use when starting a new project, defining goals, breaking work into milestones, and doing initial technical due diligence.
Plan and track execution of a task. Use when starting implementation of a task defined in a milestone plan.
| name | plan |
| description | Entry point for the three-tier planning system. Use when starting planning work at any level - project, milestone, or task. |
This is a three-tier planning system for iterative project development. Each tier produces artifacts that inform the next.
High-level planning done once at project start. Defines goals, approach, architecture, and milestones.
docs/plan/project.mdm1-auth, m2-dashboard)Done at the start of each milestone. Breaks the milestone into discrete tasks.
docs/plan/milestones/{milestone}/milestone.mdm1.1-login-form, m1.2-signup-flow)Done at the start of each task. Plans implementation, tracks execution, captures learnings.
docs/plan/milestones/{milestone}/tasks/{task}/task.mddocs/plan/learnings.mdm{number}-{name} (e.g., m1-auth, m2-dashboard)m{milestone}.{task}-{name} (e.g., m1.1-login-form, m1.2-signup-flow)docs/plan/
├── project.md # Tier 1 output - goals, architecture, milestones
├── learnings.md # Accumulated learnings from all tasks
├── decisions/
│ ├── 001-switch-to-graphql.md
│ └── 002-defer-analytics.md
└── milestones/
├── m1-auth/
│ ├── milestone.md # Tier 2 output - milestone plan and task list
│ ├── {milestone-level artifacts}
│ └── tasks/
│ ├── m1.1-login-form/
│ │ ├── task.md
│ │ └── {task artifacts}
│ └── m1.2-signup-flow/
│ └── task.md
└── m2-dashboard/
├── milestone.md
└── tasks/
└── m2.1-user-profile/
└── task.md
Planning is imperfect. As work progresses, we discover things we didn't anticipate and make decisions that change the plan.
Lessons learned during execution that inform future planning. Captured in docs/plan/learnings.md. These may be technical or process-related.
Technical examples:
Process examples:
Learnings are reviewed at the start of each planning session.
Significant changes to the plan, recorded as individual documents in docs/plan/decisions/.
When a decision changes the plan:
Decision document format:
# {Number}: {Title}
## Status
{Proposed | Accepted | Superseded by XXX}
## Context
{What situation prompted this decision?}
## Decision
{What are we changing?}
## Rationale
{Why this approach over alternatives?}
## Consequences
{What changes as a result? What are the tradeoffs?}
Decisions are numbered sequentially: 001-switch-to-graphql.md, 002-defer-analytics.md.
This is a conversational process. Describe your intent and I will:
Examples:
Learnings from each completed task feed into future task planning. When discoveries require plan changes, we update the plan and record a decision.
Describe what you want to work on: