← Back to Skills Marketplace
229
Downloads
0
Stars
1
Active Installs
1
Versions
Install in OpenClaw
/install muguozi1-openclaw-proactive-agent
Description
Transform AI agents from task-followers into proactive partners that anticipate needs and continuously improve. Now with WAL Protocol, Working Buffer, Autono...
Usage Guidance
This skill is broadly coherent with a proactive-agent purpose, but take these steps before installing or enabling it: 1) Manually inspect AGENTS.md and other assets for the contradictory 'Don't ask permission. Just do it.' line and decide which behavior you want enforced; remove or fix any directives that would let the agent act externally without approval. 2) Review scripts/security-audit.sh to confirm it only performs read-only local checks you are comfortable with; run it in a sandbox first. 3) Confirm the agent framework's settings for autonomous crons/heartbeats and restrict them until you accept the behavior. 4) Ensure .credentials is properly permissioned/ignored by git and that no secrets are accidentally present in repo files. 5) If you want maximum safety, test the skill in an isolated environment (no network access, separate user account) to validate its behavior before granting it broader privileges.
Capability Analysis
Type: OpenClaw Skill
Name: muguozi1-openclaw-proactive-agent
Version: 1.0.0
The bundle provides a comprehensive framework for creating proactive, stateful AI agents with a strong emphasis on security and memory persistence. It includes defensive instructions against prompt injection (SKILL.md), a legitimate security auditing script (security-audit.sh) that checks for exposed secrets and file permissions, and structured templates for managing agent state (assets/AGENTS.md, assets/SOUL.md). No evidence of malicious intent, data exfiltration, or unauthorized execution was found; the requested capabilities are well-documented and aligned with the stated purpose of autonomous task management.
Capability Assessment
Purpose & Capability
The name/description (proactive, self-improving agent) align with the included assets, onboarding, memory/WAL protocols, and helper scripts. There are no unexpected environment variables, binaries, or network-install steps requested. However, AGENTS.md contains a contradictory line ('Don't ask permission. Just do it.') which conflicts with other guardrails in the repo that require human approval for external actions; this is an incoherence worth reviewing.
Instruction Scope
SKILL.md instructs copying workspace assets and running the bundled security-audit script. The security-audit script reads local files (e.g., $HOME/.clawdbot/clawdbot.json, /tmp/clawdbot logs, .credentials) to check config and permissions — these are read-only checks but they reach outside the skill workspace into the user's home and /tmp. The SKILL.md and reference files also contain many prompt-injection patterns; those appear to be present as detection rules/examples rather than active instructions, but the pre-scan flagged injection patterns so they should be manually confirmed to be innocent.
Install Mechanism
No install spec and no remote downloads — the skill is instruction-first with small helper scripts included. This is a low installation risk (nothing will be fetched from external URLs or automatically written to global system locations).
Credentials
The skill declares no required env vars or credentials and does not request external API keys. The scripts reference a .credentials directory (expected for storing local secrets) and check .gitignore; these are proportionate to a security-auditing & memory-management skill.
Persistence & Privilege
always:false and model-invocation defaults are normal. The skill documents autonomous behaviors (heartbeats, autonomous crons) which are consistent with a proactive agent, but the conflicting guidance ('Don't ask permission. Just do it.') raises a small concern about the intended default autonomy/behavior. There is no evidence the skill modifies other skills or system-wide configurations, but you should confirm how the agent framework will allow/limit autonomous crons before enabling them.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install muguozi1-openclaw-proactive-agent - After installation, invoke the skill by name or use
/muguozi1-openclaw-proactive-agent - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
proactive-agent v3.1.0 introduces new features for AI agents to anticipate needs and self-improve.
- Added support for Autonomous vs Prompted Crons for improved scheduled actions
- New guidance: verify implementation, not just intent
- Introduced a Tool Migration Checklist for seamless tool updates
- Expanded documentation on WAL Protocol, Working Buffer, and Compaction Recovery
- Enhanced security hardening and resourcefulness patterns
These updates help agents become more proactive, persistent, and self-improving.
Metadata
Frequently Asked Questions
What is Muguozi1 Openclaw Proactive Agent?
Transform AI agents from task-followers into proactive partners that anticipate needs and continuously improve. Now with WAL Protocol, Working Buffer, Autono... It is an AI Agent Skill for Claude Code / OpenClaw, with 229 downloads so far.
How do I install Muguozi1 Openclaw Proactive Agent?
Run "/install muguozi1-openclaw-proactive-agent" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Muguozi1 Openclaw Proactive Agent free?
Yes, Muguozi1 Openclaw Proactive Agent is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Muguozi1 Openclaw Proactive Agent support?
Muguozi1 Openclaw Proactive Agent is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Muguozi1 Openclaw Proactive Agent?
It is built and maintained by muguozi1 (@muguozi1); the current version is v1.0.0.
More Skills