session-collab

Classifies conversation rounds into 1xx-5xx phases and manages session notes, handoffs, and sub-agent dispatch.

Updated Jun 8, 2026
One-click install
npx skills add https://github.com/wildlily1021/XW_Software --skill session-collab-wildlily1021
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: session-collab
Source: https://github.com/wildlily1021/XW_Software/tree/main/.agents/skills/session-collab
Command: npx skills add https://github.com/wildlily1021/XW_Software --skill session-collab-wildlily1021

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Long-running AI-assisted engineering conversations lose context, mix discussion with implementation, and produce duplicated or orphaned notes. This Skill enforces a structured session workflow so every non-trivial conversation is classified, documented, and handed off consistently. ## Core Features & Use Cases - Phase Classification: Routes each conversation round into 1xx (discussion), 2xx (design), 3xx (implementation), 4xx (review), or 5xx (documentation) phases with explicit entry gates. - Session Notes & Handoff: Manages topic folders under .sessions/ with session notes, topic indexes, handoff documents, and archival rules, plus deduplication checks before writing. - Sub-Agent Orchestration: Dispatches sub-agents for code reading and file scanning with a minimal prompt structure, while the main thread acts as orchestrator and verifier. - Use Case: A team running multi-session feature design can rely on the Skill to block premature implementation, require feasibility checks before design, and auto-generate handoff documents when a conversation needs to split. ## Quick Start Start a new non-trivial conversation and ask the assistant to run the session-collab entry classification to determine the current phase, required readings, and next-round plan.

Frequently Asked Questions about session-collab

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I manage context across multiple AI chat sessions?▼

Use session notes stored in topic folders under .sessions/ with a topic index and handoff documents. Each note records decisions, open items, and context so the next session inherits state without re-reading entire histories.

How to structure AI-assisted design and implementation phases?▼

Classify each round into 1xx discussion, 2xx design, 3xx implementation, 4xx review, or 5xx documentation. Entry into design requires resolved open items, feasibility checks, and coverage questions; implementation requires locked boundaries and no open disagreements.

When should a conversation be split into multiple sessions?▼

Split when the topic shifts to a new direction, roles conflict (discussing and implementing simultaneously), context saturates, or the user requests it. Do not split while convergence is ongoing or strong dependencies remain unresolved.

Can sub-agents handle code reading tasks automatically?▼

Yes, code reading, file scanning, and fact-finding tasks default to sub-agents with a minimal prompt structure covering goal, scope, constraints, deliverable, and context. The main thread verifies outputs instead of trusting them blindly.

What are the limitations of phase-gated conversation workflows?▼

The workflow adds overhead for trivial conversations and requires discipline in writing notes and handoffs. It is designed for non-trivial, multi-session engineering work and may be excessive for quick one-off questions.