ClawKit Logo
ClawKitReliability Toolkit
Back to Registry
Official Verified

nextsteps

Append context-aware next-step suggestions after agent responses. Generates actionable follow-ups, surfaces unfinished tasks from memory, and includes creative lateral suggestions. Self-learns user preferences for count, categories, and format. Activate by: (1) any response when nextsteps is enabled, (2) user says /nextsteps, "next steps", "what should I do", "suggestions", "what now", (3) user wants to customize settings, (4) user says "disable next steps" or "enable next steps".

skill-install — Terminal

Install via CLI (Recommended)

clawhub install openclaw/skills/skills/cindulasai/nextsteps
Or

NextSteps

How It Works

This skill generates helpful next-step suggestions after responses. It reads user preferences from .nextsteps/PREFERENCES.md — if enabled: false, nothing is shown. When enabled, it produces exactly display-count suggestions following the pipeline below.

Quick Start

  1. Check .nextsteps/PREFERENCES.md for user config — if enabled: false, stop (show nothing)
  2. If .nextsteps/ does not exist, run the cold-start protocol (see COLD-START.md)
  3. Generate next steps following the pipeline below
  4. Append to your response in the correct format

Generation Pipeline

Follow these six steps to generate next steps:

Step 1 — Read Configuration

Read .nextsteps/PREFERENCES.md. Extract: enabled, display-count, preferred-categories, excluded-categories, format, show-footer. Category names in PREFERENCES.md use kebab-case (e.g., direct-follow-up). If the file is missing or unreadable, use defaults: enabled=true, display-count=5, format=standard.

Step 2 — Analyze Context

Determine: What did the user just accomplish or ask? What is the active topic? What is the session scope — quick fix, feature work, exploration, debugging, architecture? What is the user's likely next thought based on conversation trajectory?

Step 3 — Check Memory

Read .nextsteps/BACKLOG.md for unfinished items relevant to current context. Read PREFERENCES.md for topic affinities (STRONG/MODERATE/WEAK) and anti-preferences (topics and types to avoid). If memory files are unavailable, skip to Step 4 — memory is a bonus, not a gate.

Step 4 — Generate Candidates

Generate display-count suggestions using these six categories:

IconCategoryTierSlot Rule
Direct Follow-upSTRONG1 guaranteed
🔧Actionable TaskSTRONG1 guaranteed
🔍Deep DiveMODERATE1 when count ≥ 3
📋Memory RecallMODERATE1 when relevant backlog exists
💡Lateral / Out-of-the-BoxMODERATE1 when count ≥ 3
Quick WinMODERATEFills remaining slots

STRONG categories get guaranteed slots. Remaining slots filled by MODERATE categories in round-robin. Respect excluded-categories. Prioritize preferred-categories. See CATEGORIES.md for detailed taxonomy and examples.

Step 5 — Self-Review Gate (CRITICAL)

Review every candidate against these core rules. Remove violators and regenerate:

  1. No restating the obvious — don't echo what was just explained
  2. No generic filler — ban "tell me more", "pros and cons?", "anything else?"
  3. No hallucinated context — only reference things the user actually mentioned
  4. No scope mismatches — match suggestion effort to session scope

Metadata

Stars3562
Views0
Updated2026-03-29
View Author Profile
AI Skill Finder

Not sure this is the right skill?

Describe what you want to build — we'll match you to the best skill from 16,000+ options.

Find the right skill
Add to Configuration

Paste this into your clawhub.json to enable this plugin.

{
  "plugins": {
    "official-cindulasai-nextsteps": {
      "enabled": true,
      "auto_update": true
    }
  }
}
Safety NoteClawKit audits metadata but not runtime behavior. Use with caution.