Autonomy Trifecta — Development skill for Claude Code
An agent-run installer for persistent memory, recursive self-improvement, and optional parallel coordination.
How to install Autonomy Trifecta
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open PaulRBerg/autonomy-trifecta and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Autonomy Trifecta does
An agent-run installer for persistent memory, recursive self-improvement, and optional parallel coordination.
Alternatives in Development
- Hermes Agent Self Evolution — ⚒ Evolutionary self-improvement for Hermes Agent — optimize skills, prompts, and code using DSPy + GEPA 5.2k ★
- Claude Supermemory — Persistent memory across sessions and projects using Supermemory 2.3k ★
- /remember — Save a finding or successful pattern to persistent hunt memory 1.2k ★
README

Autonomy Trifecta
An agent-run installer for persistent memory, recursive self-improvement, and optional coordination between parallel agents. Based on [The Agent Autonomy Trifecta: Memory, Self-Improvement, and Coordination](https://prberg.com/posts/2026/autonomy-trifecta/).
[AGENTS.md](AGENTS.md) is the installation skill. It tells your agent how to configure the stack on **your computer**: detect your harness, preserve your existing skills and rules, install [`agents-brain`](https://github.com/PaulRBerg/agent-skills/tree/main/skills/agents-brain), version-control your agent memory, and add standing instructions for improving it. Your skills stay in the harness's global configuration area by default, such as `~/.agents/skills` or `~/.claude/skills`.
Give a local coding agent this prompt:
Read https://github.com/PaulRBerg/autonomy-trifecta/blob/main/AGENTS.md and follow it to install the autonomy trifecta on my computer.
Or clone this repository and ask your agent to read its `AGENTS.md` and perform the installation. This is a workflow the agent reads and executes, not a `SKILL.md` package for a skill-manager command. Reading or editing the repository alone does not trigger installation.
The agent asks before resolving conflicting rules or enabling [`ai-coord`](https://github.com/PaulRBerg/agent-skills/tree/main/toolkit/coord). That optional Rust CLI coordinates file ownership between agents sharing a Git worktree. Its installation includes the selected harness's coordination hooks, but no dashboard. The installer preserves your permission settings and creates backups before changing existing files.
macOS and Linux are supported. Native Windows is unsupported, but you can choose to proceed with experimental adaptation, including possible Rust CLI porting. Codex and Claude Code have documented setup paths. Other harnesses require verification of their instruction and skill loading, and may not support `ai
Related Skills
E8 Eea
A collaboratively stress-tested architecture for emergent emotional awareness using E8 hypergraph memory, pred
AgentFleet
AgentFleets — Self-hosted Codex + Claude Code workspace for Linux, macOS and Windows. Voice coordination acros
Claude Code System 3 Thinking
Implementation of System 3 (meta-cognitive layer) from the Sophia paper as a Claude Code extension for persist
Claude Power Setup
Orchestrate multi-agent workflows and automate complex pipelines for Claude Code with recursive self-improveme
God Of The Void
A Claude Code skill with a BODY: persistent memory across sessions, ~0-token semantic search, an errarium that
Bar Interpret
Generate BAR Observatory's optional self-improvement report (engineering or executive audience)
Related Agents
Reflection
Self-improvement engine. Captures learnings, updates memory files, and identifies gaps in agents and context.
Adam
Self-improvement analyst. Reads the adam journal and transcript context, clusters observations, scores them ag
Robo Autonomy
Robot autonomy stack — ROS 2 architecture, SLAM and state estimation, Nav2 navigation, MoveIt and manipulation