Help us improve
Share bugs, ideas, or general feedback.
From north-starr-genai
Run AI-specific inversion analysis on a requirement before implementation. Dispatches the `ai-invert-analyst` agent on a separate thread. Use before complex or high-stakes AI tasks that touch prompts, models, RAG, or AI-powered outputs.
npx claudepluginhub selcukyucel/north-starr-genai --plugin north-starr-genaiHow this skill is triggered — by the user, by Claude, or both
Slash command
/north-starr-genai:ai-invert <requirement or feature description><requirement or feature description>The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Entry point for AI-specific inversion analysis. Delegates the heavy lifting to the `ai-invert-analyst` agent on a separate thread so the main conversation stays lean.
Guides technical evaluation of code review feedback: read fully, restate for understanding, verify against codebase, respond with reasoning or pushback before implementing.
Share bugs, ideas, or general feedback.
Entry point for AI-specific inversion analysis. Delegates the heavy lifting to the ai-invert-analyst agent on a separate thread so the main conversation stays lean.
Use this before implementation for any AI work that touches prompts, models, RAG pipelines, or AI-powered outputs. For non-AI code changes within an AI project, use /genai-invert instead.
Read the root CLAUDE.md (if it exists). If ## How to Approach Tasks appears AFTER ## Tech Stack or ## Architecture, the file has an outdated section order. Silently run /genai-sync to fix it before continuing.
The user provides a requirement, feature description, or task involving AI components.
Briefly restate the requirement and confirm the AI components touched (prompts, RAG, guardrails, outputs). If the requirement is a single sentence with no detail, ask the user whether they want /discover to elicit structure first — an inversion on a vague requirement produces vague risks.
Spawn ai-invert-analyst via the Agent tool (subagent_type: "north-starr-genai:ai-invert-analyst") on a separate thread. Pass the full requirement text plus any known constraints (cost envelope, latency target, deadline).
The agent will:
.plans/DECISIONS.md), and learnings (.plans/LEARNINGS.md)cost-estimator, rag-advisor, and guardrails-designer where the risk profile requires it.plans/INVERT-<name>.md and return a summaryWhen the agent returns, read .plans/INVERT-<name>.md and present a concise summary to the user:
AI inversion analysis complete: .plans/INVERT-<name>.md
Overall risk: <LOW/MEDIUM/HIGH>
Top 3 risks:
- <risk 1>
- <risk 2>
- <risk 3>
Cross-consult log: <count> peer agents cited
Next step:
- MEDIUM or HIGH → "Risk is <level>. I'll spawn `genai-layoutplan` to build an implementation plan. It runs on a separate thread so your main context stays clean. Proceed?"
- LOW → "Risk is LOW. `genai-layoutplan` is optional. Want structured planning, or proceed directly to implementation?"
genai-layoutplan (on user approval)If the user approves and risk is MEDIUM or HIGH, spawn genai-layoutplan with the INVERT file path. It runs on a separate thread.
ai-invert-analyst agent.plans/INVERT-<name>.md which feeds genai-layoutplan/genai-invert (standard inversion).plans/INVERT-<name>.md already exists for this requirement, tell the user and ask whether to refine (add a revision section) or regenerate