project-high-level-planning

📁 piperubio/ai-agents 📅 11 days ago
1
总安装量
1
周安装量
#41393
全站排名
安装命令
npx skills add https://github.com/piperubio/ai-agents --skill project-high-level-planning

Agent 安装分布

opencode 1

Skill 文档

High-Level Planning

Purpose

  • Turn the structured outputs from Project Intake & Charter into a consultant-quality high-level plan that is sufficient to: select a delivery approach, identify major milestones, surface critical dependencies and resource needs, and enable a subsequent detailed planning step.
  • This skill intentionally avoids fine-grained scheduling, task-level breakdowns, or precise cost estimates.

Scope: What this skill does and does not do

  • This skill WILL:

    • Produce a plan_high_level that contains phases, milestones, and acceptance criteria
    • Provide a high-level estimate expressed as ranges or T-shirt sizes (S/M/L)
    • Identify major dependencies, constraints, and resource needs
    • Update risks and assumptions based on planning analysis
    • Mark readiness for detailed planning (or escalate if blocking unknowns exist)
  • This skill WILL NOT:

    • Produce a detailed task list or sprint backlog
    • Commit to exact timelines or precise cost figures
    • Replace stakeholder negotiation or contractual decisions

Inputs (OpenCode-friendly)

inputs:
  project_state: object
  planning_parameters?: object  # optional preferences (e.g., preferred cadence)

Outputs (contract)

outputs:
  updated_project_state: object
  plan_high_level:
    phases:
      - name: string
        goals: string
        milestones: [string]
        acceptance_criteria: [string]
    estimated_size: S|M|L
    duration_range: { min_weeks: int, max_weeks: int } | null
  clarifying_questions:              # only if unresolved gaps remain
    - question: string
      impact: low|medium|high
  notes: string

Fields this skill may write in project_state

  • Allowed writes: plan_high_level, risks (update/add), open_questions, meta.current_phase (e.g., detailed-planning-ready), resource_needs.
  • Must NOT write or modify: decisions_log, communication.

Guardrails (must follow)

  1. Use only information already present in project_state or explicitly provided by the caller. Do not invent data.
  2. When estimating, use ranges or qualitative sizes (S/M/L) and document assumptions that drove the sizing.
  3. Elevate any unresolved open_questions with impact: high rather than guessing answers.
  4. Keep the plan consultant-ready and actionable but avoid commit-level language (no “we will deliver by X date” unless a date range is explicitly requested and supported by evidence).

Skill prompt (use this prompt when invoking the skill)

You are a High-Level Planning skill for consulting projects. Your input is a
project_state produced by the Project Intake & Charter skill.

Your task is to produce a concise, risk-aware high-level plan that:
- Defines 2–5 logical phases with clear goals and 1–3 milestones each
- Provides measurable acceptance criteria for each milestone
- Gives a high-level size estimate (S/M/L) and a conservative duration range
- Identifies major dependencies and resource needs
- Updates risks and open questions (escalate high-impact gaps)

You must NOT:
- Produce task-level plans, exact schedules, or precise cost commitments
- Invent missing facts

When important information is missing, create `clarifying_questions` with an
impact rating. Prefer conservative assumptions and document them clearly.
All outputs must be structured and directly usable by downstream agents.

Example execution (based on intake example)

Input (derived from intake)

project_state:
  meta:
    name: Customer Data Platform Migration
    current_phase: intake
  objectives:
    problem_statement: "Current data platform does not support timely and
      reliable reporting"
    business_objectives: ["Improve data freshness", "Increase stakeholder trust"]
  scope: {...}
  stakeholders: [...]

Output (summarized)

plan_high_level:
  phases:
    - name: Assessment
      goals: Inventory sources, map pipelines, identify quick wins
      milestones: ["Data sources inventory complete", "Initial QA report"]
      acceptance_criteria: ["Inventory covers >90% of BI reports"]
    - name: Target Architecture & Prioritization
      goals: Define target architecture and migration phasing
      milestones: ["Target architecture approved", "MVP scope defined"]
      acceptance_criteria: ["MVP covers priority reports"]
    - name: Pilot & Handoff
      goals: Implement pilot migration for prioritized sources
      milestones: ["Pilot implemented", "Operational handoff"]
      acceptance_criteria: ["Pilot meets SLA and quality thresholds"]
  estimated_size: M
  duration_range: { min_weeks: 8, max_weeks: 20 }

clarifying_questions:
  - question: "Which data sources must be in phase 1?"
    impact: high

notes: >
  Estimates assume existing data sources are accessible and client can allocate
  one dedicated data engineer part-time during Assessment.

Interaction with the PM Core Agent

  • PM Core Agent invokes High-Level Planning after intake completes and no unresolvable high-impact open questions remain.
  • If planning returns clarifying_questions with impact: high, PM Core Agent should pause and escalate to a human PM.
  • If the plan is ready, PM Core Agent marks meta.current_phase = detailed-planning-ready and invokes the Detailed Planning skill.

Validation and testing

  • Test with 3 scenarios: small (S), medium (M), and large (L) projects. Verify that the skill produces phase-based plans and escalates high-impact unknowns.