ClawKit Logo
ClawKitReliability Toolkit
Back to Registry
Official Verified

pre-task-checklist

Mandatory pre-task verification protocol to prevent forgetfulness, stale data errors, and context loss. Use when: (1) starting any non-trivial task, (2) before using stored credentials/IPs/paths, (3) when context feels uncertain, (4) after context window compaction, (5) any mention of "verify", "check", "confirm", "did I remember". Triggers on: "before I start", "pre-task", "verify this", "check if I remembered", "am I missing something", "did I forget".

skill-install — Terminal

Install via CLI (Recommended)

clawhub install openclaw/skills/skills/bowen31337/pre-task-checklist
Or

Pre-Task Checklist

Never guess. Never assume. Always verify.

The Problem

Agents forget things when:

  • MEMORY.md has stale data (old IPs, wrong paths, expired tokens)
  • Context window compaction loses recent details
  • Multiple concurrent sub-agents fragment attention
  • New tasks override previous context

Result: Wrong IPs, incorrect formats, missed rules, repeated mistakes.

The Solution

Mandatory pre-task checklist before ANY non-trivial task:

1. memory_search() for relevant patterns
2. Read reference files directly (don't assume)
3. Verify critical details in real-time
4. Check AGENTS.md / SOUL.md for rules

Quick Start

# Before starting a task
source ~/.openclaw/workspace/skills/pre-task-checklist/scripts/checklist.sh

# Verify a specific detail (IP, path, format)
verify "GPU server IP"
verify "MBD article format"

# Full checklist (interactive)
run_checklist

Checklist Items

1. Memory Search

# Search for relevant patterns
memory_search "GPU server IP address"
memory_search "MBD article format"
memory_search "folder naming convention"

What to look for:

  • Previous examples of this task
  • Correct IPs, paths, formats
  • Rules or constraints
  • Past mistakes to avoid

2. Read Reference Files

# Read actual examples, don't assume
cat /tmp/mbd-book-ideas/2026-03-11_懒人理财经_AI替你管钱躺赚不是梦/文章.md
cat ~/.openclaw/workspace/TOOLS.md
cat ~/.openclaw/workspace/AGENTS.md

What to verify:

  • Format and structure
  • Tone and style
  • Technical requirements
  • Examples to follow

3. Verify Critical Details

# Check live data, don't trust memory
ssh [email protected] "echo 'connected'"  # Verify IP works
ls -la /path/to/file                     # Verify path exists
cat ~/.openclaw/workspace/TOOLS.md       # Verify current value

What to verify:

  • IPs and hostnames
  • File paths and URLs
  • API keys and tokens (check expiry)
  • Configuration values

4. Check Rules

# Read rule files before acting
cat ~/.openclaw/workspace/AGENTS.md      # Check project rules
cat ~/.openclaw/workspace/SOUL.md        # Check identity/rules
cat ~/.openclaw/workspace/HEARTBEAT.md   # Check recurring tasks

What to check:

  • "No pause" rule - execute immediately
  • Repo ownership (clawinfra vs AlexChen31337)
  • Publish gates (MbD needs explicit approval)
  • Safety/security rules

Examples

Example 1: GPU Server Access

Wrong (guessing):

ssh [email protected]  # WRONG IP from stale memory

Right (verifying):

# Step 1: Search memory
memory_search "GPU server IP"
# → Returns: "10.0.0.44" (STALE!)

# Step 2: Verify in TOOLS.md
cat ~/.openclaw/workspace/TOOLS.md | grep "GPU Server"
# → Shows: "[email protected]" (CURRENT!)

# Step 3: Test connection
ssh [email protected] "hostname"
# → Works!

# Step 4: Update memory
# (Correct the stale data)

Metadata

Stars4190
Views1
Updated2026-04-18
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-bowen31337-pre-task-checklist": {
      "enabled": true,
      "auto_update": true
    }
  }
}
Safety NoteClawKit audits metadata but not runtime behavior. Use with caution.

Related Skills

Terse

Skill by bowen31337

bowen31337 4190

Identity Resolver

Skill by bowen31337

bowen31337 4190

agent-self-governance

Self-governance protocol for autonomous agents: WAL (Write-Ahead Log), VBR (Verify Before Reporting), ADL (Anti-Divergence Limit), VFM (Value-For-Money), and IKL (Infrastructure Knowledge Logging). Use when: (1) receiving a user correction — log it before responding, (2) making an important decision or analysis — log it before continuing, (3) pre-compaction memory flush — flush the working buffer to WAL, (4) session start — replay unapplied WAL entries to restore lost context, (5) any time you want to ensure something survives compaction, (6) before claiming a task is done — verify it, (7) periodic self-check — am I drifting from my persona? (8) cost tracking — was that expensive operation worth it? (9) discovering infrastructure — log hardware/service specs immediately.

bowen31337 4190

rsi-loop

Recursive Self-Improvement (RSI) loop for EvoClaw agents. Provides a structured observe→analyze→synthesize→deploy pipeline that enables agents to detect their own failure patterns and generate concrete improvement proposals (new skills, routing fixes, SOUL.md updates, memory improvements). Use when: (1) logging a task outcome (success/fail/quality), (2) running periodic self-improvement analysis, (3) reviewing or deploying improvement proposals, (4) integrating RSI into EvoClaw hub/edge agents via MQTT, (5) checking agent health score, (6) any mention of "self-improvement", "recursive improvement", "fix my own mistakes", "improvement loop", or "agent evolution". Core EvoClaw primitive.

bowen31337 4190

clawchain

ClawChain RPC client for EvoClaw agents. Connects to Substrate-based blockchain, queries on-chain agent data, submits transactions, and enables agents to participate in on-chain governance and reputation tracking. Use when working with ClawChain L1 blockchain, agent DIDs, token economics, or agent reputation systems.

bowen31337 4190