skills.rest
good-night-oppie
good-night-oppieOFFICIAL
Explore all skills by this author →
GitHub

bene

Operates the BENE multi-agent harness for AI coding agents with durable memory and eval gates.

1|Updated May 30, 2026
Search Tags:#multi-agent#memory#coordination#evolution#autonomy#promotion#eval gates
One-click install
npx skills add https://github.com/good-night-oppie/bene --skill bene
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: bene
Source: https://github.com/good-night-oppie/bene/tree/main/site
Command: npx skills add https://github.com/good-night-oppie/bene --skill bene

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires python, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill harnesses BENE, a multi-agent framework for AI coding agents that remembers, evolves, and never starts cold, addressing the challenges of durable cross-session memory, reproducible eval gates, planner-executor or fan-out multi-agent coordination, and auditable promotion of evolved prompts/skills.

Core Features & Use Cases

  • Multi-Agent Coordination: Facilitates parallel worker, planner-executor, and fan-out multi-agent coordination.
  • Durable Memory: Offers durable cross-session memory with engrams and provenance for reproducibility.
  • Eval Gates & Promotion: Implements falsifiable eval probes with hash-locked kill gates and kill-gated promotion of evolved strategies.
  • Autonomy & Trust: Provides autonomy ladder L0-L4 and computed trust for safe agent operations.
  • Context Budgeting & Pollution Recovery: Manages context assembly and pollution recovery for efficient memory usage.
  • Evolutionary Meta-Harness Search: Enables model evolution and meta-harness search for continuous improvement.
  • MCP Server: Integrates with 37 tools for agent operations.
  • Use Case: Ideal for building/orchestrating agent apps, services, pipelines, or content systems that require robust agent memory, reproducible eval gates, and auditable promotion.

Quick Start

Initialize BENE with bene init and run a demo with bene demo --no-ui.

Frequently Asked Questions about bene

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do multi-agent coordination frameworks handle parallel worker execution?▼

Falsifiable eval probes validate evolved prompts using hash-locked kill gates. This kill-gated promotion mechanism only allows agent strategies to advance when they pass strict, auditable evaluation criteria.

What is the best way to prevent context pollution in autonomous AI agents?▼

Multi-agent coordination supports parallel worker, planner-executor, and fan-out execution patterns. This allows AI coding agents to distribute tasks efficiently while maintaining computed trust through an autonomy ladder ranging from L0 to L4.

Do I need Python and SQLite to operate a multi-agent harness with MCP server integration?▼

Context pollution recovery and context budgeting manage memory assembly to prevent overload. This approach efficiently allocates context windows, ensuring autonomous agents operate within safe boundaries during complex multi-agent harness operations.

How do I initialize and test a multi-agent harness for AI coding agents?▼

Yes, operating this multi-agent harness requires Python 3.11+ and SQLite. The environment supports a SQLite-based virtual filesystem and integrates an MCP server exposing 37 tools for comprehensive agent operations.

When should I use evolutionary meta-harness search for AI agent development?▼

Initialize the harness using the `bene init` command, then execute a demonstration via `bene demo --no-ui`. This validates the durable memory, eval gates, and planner-executor coordination setup without requiring a UI.

When should I use evolutionary meta-harness search for AI agent development?▼

Use evolutionary meta-harness search when you need continuous model improvement and strategy evolution. It enables the system to automatically discover and promote better agent configurations through falsifiable probes and auditable promotion.

Related Skills

Explore More Software Engineering Skills→
officechbusinessservices-creatorofficechbusinessservices-creator

mcp-setup

Configure MCP servers via claude mcp add to extend Claude Code agents.

Community
Advanced
IoT-VNIoT-VN

mcp-management

Manage MCP servers for tool discovery and execution via subagents.

Community
Advanced
skyllwtskyllwt
1.6k

/init

Orchestrate ΩmegaWiki initialization and parallel paper ingestion from raw sources.

Community
Advanced
northseadlnorthseadl

memory-systems

Implements agent memory systems for AI assistants to persist state across sessions and build knowledge graphs.

Community
Advanced