- Home
- Skills
- Catlog22
- Claude Code Workflow
- Collaborative Plan With File
collaborative-plan-with-file_skill
- TypeScript
1.3k
GitHub Stars
1
Bundled Files
3 weeks ago
Catalog Refreshed
2 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstart where the catalogue uses aiagentskills.
npx veilstart add skill catlog22/claude-code-workflow --skill collaborative-plan-with-file- SKILL.md31.5 KB
Overview
This skill implements a serial collaborative planning workflow centered on a shared Plan Note (plan-note.md). It analyzes requirements, splits work into 2–5 sub-domains, generates per-domain task JSON, and runs a single-source conflict detection pass. The process produces a unified .task/ set, conflicts.json, and a human-readable plan.md without merge headaches or agent delegation.
How this skill works
The skill first analyzes the request and project context to identify sub-domains and create a plan-note.md template with pre-allocated sections and non-overlapping TASK ID ranges. It then plans each domain serially: exploring the codebase, producing per-domain .task/TASK-*.json files, and appending evidence to plan-note.md. After all domains finish, it parses plan-note.md to detect file, dependency, and strategy conflicts, writes conflicts.json, merges task files into a session .task/ folder, and generates plan.md for execution.
When to use it
- When you need a structured, auditable plan across multiple technical domains without parallel merge conflicts.
- For medium-to-large feature work that benefits from domain-focused, sequential analysis (backend, frontend, infra, testing, etc.).
- When you want automated conflict detection across tasks before execution.
- If you need machine-readable task artifacts (.task/*.json) plus a human-readable summary (plan.md).
Best practices
- Provide a clear, concise original requirement to improve sub-domain identification accuracy.
- Keep max domains reasonable (default 5) to maintain focused, high-quality domain plans.
- Review plan-note.md evidence sections after each domain to catch hidden dependencies early.
- Use auto mode (--yes or -y) for unattended runs, but confirm splits manually for ambiguous requirements.
- Reference project READMEs and architecture docs in the working directory to improve search results.
Example use cases
- Plan a real-time notification system: identify API, worker, storage, delivery, and testing domains and produce a unified task set.
- Refactor authentication: split into token handling, session store, login UI, and tests, then detect overlapping file changes.
- Add payment gateway support: generate domain-specific tasks for payments API, UI flows, security checks, and infra configuration.
- Produce a pre-execution plan for large features so CI/CD teams can review tasks, dependencies, and conflict reports.
FAQ
No. Planning is serial per domain and runs in a single source of truth (plan-note.md). There is no agent delegation; each domain is planned sequentially with full context.
How are TASK IDs assigned to avoid collisions?
Each domain receives a non-overlapping 100-ID range (e.g., TASK-001–100, TASK-101–200) allocated in the plan-note.md template so ranges never overlap.