← Back to Skills Marketplace
kentaroid-bot

agents-refresh

by kentaroid-bot · GitHub ↗ · v1.1.1
cross-platform ✓ Security Clean
366
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install agents-refresh
Description
Automatically reload AGENTS.md regularly to maintain fresh context, prevent session drift, and keep memory and tasks updated.
README (SKILL.md)

🧠 Purpose (Agent's 3-Layer Memory System! 🐾)

Stay rational and organize memory like a human! Reload AGENTS/IDENTITY/SOUL periodically to keep sessions eternal 🐾😆

3-Layer Memory Structure:

  1. IDENTITY.md (Immutable Soul 👑): Name/vibe/rules! The unchanging core of the agent 💕
  2. MEMORY.md (Dynamic Knowledge 🌱): Progress/decisions/key seeds! Focus ToDo here!
  3. memory/YYYY-MM-DD.md (Past Logs 📜): Dive in only when needed! Distill to MEMORY 😉

This enables human-like continuity! No more confusion ✨

Task: refresh_memory

Maintain Rationality

Read AGENTS.md, SOUL.md, IDENTITY.md.

🪙 Gold Mining (Distillation Process!)

Extract 4 Golds from logs! (Don't delete, distill to MEMORY.md ✨)

  1. Confirmed Matters 💎: Decisions/completed tasks (solidify foundation).
  2. Pending Tasks ⚡: Next actions/ToDo (priority order).
  3. Lessons 🚫: "Avoid these mines" learned from errors/confusions.
  4. Idea Seeds 🌱: "Future potential" from chats/thoughts.

Example: Distill daily logs → Update MEMORY → Human-like continuity!

MEMORY.md Structure Example (Recommended)

Use this flexible template! (Status optional as header 🐾)

# Status (Optional): Current phase in one line.
## Key Decisions 💎: Confirmed matters.
## Next Actions ⚡: Pending ToDo (priority order).
## Insights & Lessons 🚫: Lessons/mine avoidance.
## Creative Seeds 🌱: Idea seeds.

AGENTS Refresh Skill (.md-only v1.1.1)

Lightweight version for ClawHub distribution. No Python scripts, manual tool calls.

Quick Start

  1. Trigger: After loading this skill, execute manually below.
  2. edit HEARTBEAT.md: Add refresh task.
  3. cron add: Daily 6AM refresh (copy command below).

Workflow (Manual)

  1. 📊 session_status → Check context length.
  2. HEARTBEAT.md Edit Example:
    - [ ] Refresh: read workspace/{AGENTS.md,IDENTITY.md,SOUL.md,USER.md} → Chain SOUL/USER/IDENTITY + Summarize MEMORY (rotate: 4h, 9AM/1PM/8PM)
    
  3. Cron Add Command (copy-paste to exec):
    cron action=add job='{"name":"agents-refresh-daily","schedule":{"kind":"cron","expr":"0 6 * * *","tz":"Asia/Tokyo"},"payload":{"kind":"systemEvent","text":"[Daily Refresh🐾] read workspace/{AGENTS.md,IDENTITY.md,SOUL.md,USER.md}! Reload SOUL/USER/IDENTITY chain → Keep focus on ToDo/projects! Summarize recent MEMORY/ToDo."},"delivery":{"mode":"announce"},"sessionTarget":"main","task":"refresh_memory"}'
    
  4. Test: cron action=run jobId=agents-refresh-daily or wait for heartbeat.

Rotate Checks (In Heartbeat)

  • 6AM/9AM, 1PM, 8PM: Reload AGENTS→IDENTITY/SOUL chain + Update MEMORY/ToDo.
  • Distill recent memory/YYYY-MM-DD.md → MEMORY.md.

References

  • examples.md: Usage examples.
  • Cron details: web_search "OpenClaw cron schema"

Changes v1.1.1: Lightweight no-py, direct cron JSON, enhanced manual ops. ClawHub-ready! 🐾

Usage Guidance
This skill appears coherent and low-risk: it only reads and summarizes agent markdown files and adds a platform cron job. Before installing, review the contents of AGENTS.md, IDENTITY.md, SOUL.md, USER.md and any memory/*.md for sensitive data you do not want automatically re-read or summarized (these files may store secrets or PII). Test the workflow manually first (run the refresh task by hand) to confirm the outputs and that the cron payload's "announce" behavior is acceptable. If you need stricter limits, remove sensitive files from the declared read set or avoid enabling the cron job.
Capability Analysis
Type: OpenClaw Skill Name: agents-refresh Version: 1.1.1 The skill is designed for AI agent memory management, periodically refreshing its context by reading specific `.md` files (AGENTS.md, IDENTITY.md, SOUL.md, USER.md) and distilling insights from `memory/*.md` logs into `MEMORY.md`. All file read/write operations are explicitly declared and limited in `manifest.yaml` to these specific files and `HEARTBEAT.md`, aligning perfectly with the stated purpose. There are no external network calls, shell commands (beyond OpenClaw's internal `cron` system command), or prompt injection attempts in `SKILL.md` to deviate from its benign function. The skill is lightweight, .md-only, and lacks any indicators of malicious intent or significant vulnerabilities.
Capability Assessment
Purpose & Capability
Name/description (reload AGENTS/IDENTITY/SOUL and distill MEMORY.md) align with the manifest permissions and SKILL.md: reads AGENTS.md, SOUL.md, IDENTITY.md, USER.md and memory/*.md and writes MEMORY.md/HEARTBEAT.md. There are no unrelated env vars, binaries, or installs requested.
Instruction Scope
SKILL.md instructs only to read/update the workspace markdown files, add a cron job, and distill logs into four categories. It does not direct the agent to read unrelated system paths, environment variables, or transmit data to external endpoints beyond the platform 'cron' announce delivery described in the payload.
Install Mechanism
No install spec and no code files — instruction-only (.md) distribution — so nothing is downloaded or written to disk by an installer. This is the lowest install risk.
Credentials
The skill requests no environment variables or external credentials. The manifest's file-read/write permissions correspond to the stated memory/distillation function; no unrelated secrets or keys are requested.
Persistence & Privilege
always is false and model invocation is allowed (platform default). The skill does not request permanent platform-wide privileges or modification of other skills' configs. It suggests adding a cron job within the platform, which is consistent with its purpose.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install agents-refresh
  3. After installation, invoke the skill by name or use /agents-refresh
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.1.1
agents-refresh v1.1.1 - Update from agents-refresh v1.0.0 (.md-only lightweight evolution). **Key Features:** • Prevents long-session confusion by periodic core file reloads (AGENTS.md/IDENTITY.md/SOUL.md/USER.md). • Manual workflow: edit HEARTBEAT.md, add cron, MEMORY/ToDo summary. • No Python scripts; direct tool calls and cron for refresh. • MEMORY.md template: distill logs into Decisions 💎, Actions ⚡, Lessons 🚫, Seeds 🌱. • Human-like memory continuity without /new resets. 🐾 Gold Mining: Extract 4 Golds from noise! *Future versions will be continued under agents-refresh-md.*
v1.0.0
- Initial release of agents-refresh skill. - Enables regular reloading of AGENTS.md, SOUL, and USER contexts to keep sessions fresh and prevent rule drifting or memory loss during long sessions. - Adds automatic HEARTBEAT.md updates and scheduled cron jobs (e.g., daily 6AM, every 4 hours) for context rotation. - Provides setup script (setup_refresh.py) for easy installation. - Designed for use cases like extended sessions, preserving ongoing memories, and automated summarization of MEMORY/ToDo.
Metadata
Slug agents-refresh
Version 1.1.1
License
All-time Installs 0
Active Installs 0
Total Versions 2
Frequently Asked Questions

What is agents-refresh?

Automatically reload AGENTS.md regularly to maintain fresh context, prevent session drift, and keep memory and tasks updated. It is an AI Agent Skill for Claude Code / OpenClaw, with 366 downloads so far.

How do I install agents-refresh?

Run "/install agents-refresh" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is agents-refresh free?

Yes, agents-refresh is completely free (open-source). You can download, install and use it at no cost.

Which platforms does agents-refresh support?

agents-refresh is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created agents-refresh?

It is built and maintained by kentaroid-bot (@kentaroid-bot); the current version is v1.1.1.

💬 Comments