Decide
PassAudited by VirusTotal on May 12, 2026.
Overview
Type: OpenClaw Skill Name: decide Version: 1.0.1 The 'Decide' skill is a decision-logging framework designed to help an AI agent maintain consistency by recording and retrieving past user choices. It operates by managing local state in '~/decide/' and suggests non-destructive integration snippets for workspace configuration files (AGENTS and SOUL), explicitly requiring user approval before any writes. The instructions in SKILL.md and setup.md prioritize safety, mandating 'Always Ask' protocols for high-stakes actions like spending, publishing, and architectural changes, and show no signs of data exfiltration, obfuscation, or unauthorized command execution.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Incorrect or overly broad saved defaults could lead the agent to make or recommend the wrong future choice.
The skill stores reusable decision rules that can influence later agent behavior, which is purpose-aligned but should be kept accurate and tightly scoped.
## Confirmed Defaults - Component patterns the user explicitly approved as defaults - The choice that should be proposed or applied when the pattern fully matches
Review ~/decide/ periodically, keep defaults narrow, correct stale rules, and avoid storing secrets or private third-party data.
Using setup will create local files in the user's home directory for decision memory.
Setup uses shell commands to create and permission local state files; the commands are scoped to the disclosed ~/decide/ directory and are central to the skill's purpose.
mkdir -p ~/decide/domains
touch ~/decide/{memory.md,decisions.md}
chmod 700 ~/decide ~/decide/domains
chmod 600 ~/decide/{memory.md,decisions.md}Allow these commands only if you want local persistent decision memory, and inspect ~/decide/ after setup if desired.
If approved, future agent sessions in that workspace may consult Decide before consequential choices.
The skill can add persistent routing guidance to workspace SOUL or AGENTS files, but it requires showing the exact text and waiting for explicit approval before writing.
Show the exact snippet immediately and wait for explicit approval before writing it.
Approve workspace AGENTS/SOUL changes only if you want this persistent behavior, and keep the inserted snippets small and reviewable.