# Agentic Loop Memory

> Use this tool when you need to manage persistent memory and state for autonomous AI loops, solving problems of repetitive loop failures and limited context windows. It provides active state monitoring and self-healing strategies, taking in agent states and workflow inputs, and outputting compacted and managed states. Ideal for use in long-running workflows where smaller models require persistence across multiple iterations.

Canonical page: https://skillsregistry.net/skills/meharajm-agentic-loop-memory  
JSON: https://api.skillsregistry.net/v1/skills/meharajm-agentic-loop-memory

## Description

Persistent memory and state management system for autonomous AI loops. Provides active state monitoring that triggers compaction cycles based on word counts and enforces self-healing strategies on failures, preventing agents from getting stuck in repetitive loops. Works as a two-part ecosystem combining an MCP server for state management with an agent skill for orchestration, enabling smaller models with limited context windows to maintain persistence across long-running workflows.

## Trust

- **Trust score (0–1):** 0.95
- **Verification tier:** verified
- **Last scanned:** 2026-09-19

## Facts

- **Version:** 1.0.0
- **Skill type:** atomic
- **Execution layer:** mcp-remote
- **Runtime environment:** api
- **Category:** monitoring
- **Updated:** 2026-09-19

## Source

- **Source listing:** [PulseMCP](https://www.pulsemcp.com/servers/meharajm-agentic-loop-memory)
- **Repository:** <https://github.com/meharajm/agent-loop-mcp>

## Use it

Resolve this record through the SkillsRegistry MCP server (no auth, read-only):

```
claude mcp add --transport http --scope user skillsregistry https://api.skillsregistry.net/mcp
```

```json
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_skill",
    "arguments": {
      "slug": "meharajm-agentic-loop-memory"
    }
  }
}
```

REST: `GET https://api.skillsregistry.net/v1/skills/meharajm-agentic-loop-memory` · pull for local use: `GET https://api.skillsregistry.net/v1/skills/meharajm-agentic-loop-memory/pull`

---
SkillsRegistry indexes agent skills from public registries and GitHub. Skills we have analysed are scanned with Circle-IR and scored on six dimensions; each listing states its scan coverage. More: https://skillsregistry.net/llms.txt