スキル一覧に戻る
DazedtilDawn

edge-plan

by DazedtilDawn

Mechanical enforcement system for Claude Code - YAML state tracking, hook-based enforcement, Three Gears autopilot

0🍴 0📅 2026年1月21日
GitHubで見るManusで実行

SKILL.md


name: edge-plan description: Create or update a structured plan in active_context.yaml. Use when starting work on a new objective or refining an existing plan.

Create/Update Plan

Read the current state and help create or refine a plan.

Current State

First, read active_context.yaml in the project root.

Instructions

0. Check for Research Needs First

Before creating a plan, scan the objective for research needs:

Check for these signals:

  • Complex technologies mentioned (kubernetes, websocket, CRDT, etc.)
  • Ambiguity signals ("best way", "should I", "optimize", "compare")
  • Explicit research verbs ("evaluate", "investigate", "research")

If critical research needs detected:

RESEARCH RECOMMENDED BEFORE PLANNING

Detected [N] research need(s) that may block planning:

CRITICAL:
  - [topic] - [reason]

Recommendation: Run $edge-research to generate prompts
for external deep research before creating the plan.

Continue anyway? (Some decisions may need to be revisited)

If no critical research needs, proceed with planning.

1. Understand the objective

What are we trying to achieve? If no objective is set, ask the user.

2. Break it into steps

Each step should be:

  • Small enough to complete in one focused effort
  • Testable (you can verify it worked)
  • Independent where possible

3. Update active_context.yaml

objective: "Clear statement of goal"
current_step: 1
plan:
  - description: "First step"
    status: pending
    proof: null
  - description: "Second step"
    status: pending
    proof: null
constraints:
  - "What must NOT happen"
risks:
  - risk: "What could go wrong"
    mitigation: "How to prevent/detect it"

4. Identify constraints

What should NOT change? What's off-limits?

5. Identify Risks (REQUIRED)

Ask yourself: "What are 3 ways this could fail?"

Risk TypeExamples
TechnicalWrong library, API changes, version conflicts
ScopeFeature creep, unclear requirements, missing edge cases
IntegrationBreaks existing code, data format mismatch
ExternalNetwork failures, rate limits, auth issues
risks:
  - risk: "psycopg2 writes to PostgreSQL but app uses SQLite"
    mitigation: "Verify database connection string matches app config"
  - risk: "Sync might timeout on large datasets"
    mitigation: "Add batch processing with progress logging"

Tip: If you can't think of risks, the objective may be too vague.

Output

After updating active_context.yaml, summarize:

  • The objective
  • Number of steps
  • Key constraints
  • Top risks (at least 1 required)
  • Any research recommendations

Do NOT start implementation. Just plan.

Arguments

  • No arguments: Create plan for current objective, or ask for one
  • "objective text": Set this as the new objective and create plan

スコア

総合スコア

50/100

リポジトリの品質指標に基づく評価

SKILL.md

SKILL.mdファイルが含まれている

+20
LICENSE

ライセンスが設定されている

0/10
説明文

100文字以上の説明がある

+10
人気

GitHub Stars 100以上

0/15
最近の活動

3ヶ月以内に更新がある

0/10
フォーク

10回以上フォークされている

0/5
Issue管理

オープンIssueが50未満

+5
言語

プログラミング言語が設定されている

+5
タグ

1つ以上のタグが設定されている

0/5

レビュー

💬

レビュー機能は近日公開予定です