@framers/agentos 0.1.150 → 0.1.151
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +34 -2
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -6,7 +6,7 @@
|
|
|
6
6
|
|
|
7
7
|
# AgentOS
|
|
8
8
|
|
|
9
|
-
**
|
|
9
|
+
**Build autonomous AI agents with adaptive intelligence and emergent behaviors. Multimodal RAG, cognitive memory, streaming guardrails, voice pipeline, and 21 LLM providers. Open-source TypeScript runtime.**
|
|
10
10
|
|
|
11
11
|
[](https://www.npmjs.com/package/@framers/agentos)
|
|
12
12
|
[](https://github.com/framersai/agentos/actions/workflows/ci.yml)
|
|
@@ -24,6 +24,8 @@
|
|
|
24
24
|
## Table of Contents
|
|
25
25
|
|
|
26
26
|
- [Overview](#overview)
|
|
27
|
+
- [Adaptive Intelligence](#adaptive-intelligence)
|
|
28
|
+
- [Emergent Behaviors](#emergent-behaviors)
|
|
27
29
|
- [Quick Start](#quick-start)
|
|
28
30
|
- [1. Generate Text (Provider-First)](#1-generate-text-provider-first)
|
|
29
31
|
- [2. Agent with Personality & Memory](#2-agent-with-personality--memory)
|
|
@@ -91,7 +93,9 @@
|
|
|
91
93
|
|
|
92
94
|
## Overview
|
|
93
95
|
|
|
94
|
-
`@framers/agentos` is an open-source TypeScript AI agent runtime for building
|
|
96
|
+
`@framers/agentos` is an open-source TypeScript AI agent runtime for building autonomous AI agents with **adaptive intelligence** and **emergent behaviors**. Agents continuously improve through meta-reflective prompt adaptation, self-evaluating response quality, personality-modulated cognition, and autonomous memory consolidation. At runtime, agents forge new tools, adapt their personality, compose workflows, and promote reliable capabilities for cross-agent reuse.
|
|
97
|
+
|
|
98
|
+
The runtime provides multimodal RAG with cognitive memory (Ebbinghaus decay, 8 neuroscience-backed mechanisms including reconsolidation, retrieval-induced forgetting, involuntary recall, metacognitive FOK, temporal gist extraction, schema encoding, source confidence decay, and emotion regulation ��� all HEXACO personality-modulated), multi-agent orchestration, 37 channel adapters, 6 guardrail packs (PII redaction, ML classifiers, topicality, code safety, grounding guard, content policy rewriter) with prompt injection defense, 21 LLM providers, and 72 curated skills. Self-hostable and production-ready, it handles the full lifecycle from prompt construction through tool execution, safety evaluation, and streaming response delivery.
|
|
95
99
|
|
|
96
100
|
**Key facts:**
|
|
97
101
|
|
|
@@ -154,6 +158,34 @@
|
|
|
154
158
|
|
|
155
159
|
---
|
|
156
160
|
|
|
161
|
+
## Adaptive Intelligence
|
|
162
|
+
|
|
163
|
+
AgentOS agents continuously improve their behavior without retraining through five interconnected mechanisms:
|
|
164
|
+
|
|
165
|
+
| Mechanism | How It Works |
|
|
166
|
+
|-----------|-------------|
|
|
167
|
+
| **Meta-reflective prompt adaptation** | The PromptBuilder assembles a different system prompt every turn, dynamically incorporating personality traits, mood state, conversation history, retrieved memories, and available tools. No two turns use the same prompt. |
|
|
168
|
+
| **Self-evaluating response quality** | The `self_evaluate` tool scores the agent's own output and adjusts parameters (temperature, verbosity, personality expression) in real time, creating a continuous feedback loop. |
|
|
169
|
+
| **Personality-modulated cognition** | HEXACO traits modulate how the agent processes information. High openness increases creative associations during memory retrieval; high conscientiousness strengthens retrieval-induced forgetting of irrelevant data. |
|
|
170
|
+
| **Autonomous memory consolidation** | The `ConsolidationLoop` prunes weak memories, strengthens frequently-accessed ones, and derives new insights from memory clusters. The agent's knowledge base improves over time without explicit training. |
|
|
171
|
+
| **QueryRouter tiered classification** | The system adapts retrieval depth based on query complexity. Simple questions get fast keyword lookup; complex questions trigger full hybrid RAG with deep research. |
|
|
172
|
+
|
|
173
|
+
---
|
|
174
|
+
|
|
175
|
+
## Emergent Behaviors
|
|
176
|
+
|
|
177
|
+
AgentOS agents develop new capabilities at runtime rather than having them explicitly programmed:
|
|
178
|
+
|
|
179
|
+
| Capability | How It Works |
|
|
180
|
+
|-----------|-------------|
|
|
181
|
+
| **Runtime tool forging** | Agents create new tools on the fly via `forge_tool`. The `EmergentCapabilityEngine` uses sandboxed JavaScript execution and LLM-as-judge evaluation to safely create, test, and promote tools. |
|
|
182
|
+
| **Self-improving personality** | Agents adapt their HEXACO personality traits within bounded limits via `adapt_personality`. Mutations persist with Ebbinghaus decay — strong repeated adaptations stick, while weak ones fade naturally. |
|
|
183
|
+
| **Dynamic skill management** | Agents enable or disable skills at runtime via `manage_skills`, adapting their behavioral repertoire to the task at hand. |
|
|
184
|
+
| **Composable workflow creation** | Agents compose registered tools into multi-step pipelines via `create_workflow`, building new capabilities from existing building blocks. |
|
|
185
|
+
| **Tiered tool promotion** | Forged tools progress through session, agent, and shared tiers. Tools that prove reliable (5+ successful uses, >0.8 confidence score) auto-promote for cross-agent reuse. |
|
|
186
|
+
|
|
187
|
+
---
|
|
188
|
+
|
|
157
189
|
## Quick Start
|
|
158
190
|
|
|
159
191
|
```bash
|