echo-agent

What it is and what it does

Echo Agent is a framework for deploying AI agents that retain memory across conversations and continuously improve their skills from real execution traces. Unlike stateless chatbots, it maintains four layers of cognitive memory (working, episodic, semantic, archival) with automatic decay and contradiction detection to prevent memory bloat over long-term operation. Skills are generated as candidates from execution logs, validated against test cases, and promoted or rolled back—allowing the agent to improve without manual retraining.

The runtime unifies multiple entry points (CLI, HTTP gateway, webhooks, cron jobs, and 12 messaging platforms including Telegram, Discord, Slack, WeChat, Feishu, and DingTalk) into a single shared state and execution context. It enforces safety through tiered tool approval (manual, smart, or off), encrypts credentials at rest, and logs all actions for audit. Designed for self-hosted deployment on your own infrastructure, it supports local-first storage of sessions, memory, execution traces, and secrets.

Use it for:

• Run a persistent AI assistant on your server that remembers conversation history and user preferences across weeks or months, improving responses over time.
• Integrate the same agent across multiple communication channels (CLI, Slack, Discord, webhooks) while maintaining unified memory and permissions.
• Deploy an agent that automatically refines its tool-use strategies by analyzing real execution outcomes, with human approval gates for risky operations.
• Build audit-compliant AI workflows where every tool call, decision, and memory update is logged and traceable for compliance or debugging.
• Automate recurring tasks (via cron) while the agent learns from past executions and adapts its approach without code changes.
• Operate a multi-model agent that routes different tasks (reasoning, embedding, approval) to different LLM providers based on cost or capability.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Echo Agent is a self-hosted, long-running AI agent runtime that maintains cross-session memory, evolves skills from execution traces, and integrates with multiple communication channels (CLI, webhooks, messaging platforms) while enforcing tool approval and credential security.

Yes, if you need a self-hosted agent that retains memory and improves over time. The framework is actively maintained, has low install friction, and covers real gaps in stateless agent systems. Start with `echo-agent setup` to configure your model API keys. Not suitable if you need a simple one-shot chatbot or prefer fully managed cloud hosting.

Install

echo-agent on PyPI

pip install echo-agent

uv add echo-agent

poetry add echo-agent

Installing echo-agent

Before you install

Low friction: pure Python wheel with 11 runtime dependencies including aiohttp, pydantic, and loguru. Active maintenance (last commit 2026-08-14, 928 GitHub stars). Requires Python 3.11+. No compiled dependencies reported.

License in practice

MIT license (permissive) allows commercial and private use with minimal restrictions—include a copy of the license and copyright notice.

Quickstart

pip install "echo-agent[all]"
echo-agent setup
echo-agent run

Requires Python 3.11+. Setup requires at least one model API key (OpenAI, Claude, or other supported provider).

Verify before relying

• Whether the four-layer memory system (Working/Episodic/Semantic/Archival) scales efficiently for months or years of continuous operation.
• Performance characteristics of BM25 + FAISS hybrid retrieval under high-volume memory scenarios.
• Actual overhead and latency of the self-evolution engine (trajectory recording, candidate generation, evaluation).
• Whether the tool approval system's three policies (manual/smart/off) cover all real-world risk profiles.

Package facts

Evidence: echo_agent-0.3.5-py3-none-any.whl

Keywords: agent, ai, mcp, memory, plugins, runtime

Tags

More Artificial Intelligence packages

Further reading