Stats
Actions
Available In
Tags
Plugin
hyperagents
Self-referential self-improving agent framework for Claude Code. Inspired by Meta's HyperAgents paper — evolutionary optimization of skills, agents, hooks, and code via LLM-driven mutation and fitness-based selection.
What's Inside
Slash Commands5
archive
/archive
View, query, and manage the HyperAgents evolutionary archive. Shows generation history, fitness scores, lineage trees, and best performers. Use: /hyperagents:archive [show|best|lineage|prune|export]
evaluate
/evaluate
Evaluate a specific generation or the current codebase against fitness criteria. Supports staged (quick) and full evaluation modes. Use: /hyperagents:evaluate [--genid <id>] [--domain <domain>] [--staged]
evolve
/evolve
Start or resume a HyperAgents evolutionary improvement loop. Iteratively mutates code/skills/agents, evaluates fitness, and selects parents for the next generation. Use: /hyperagents:evolve [--domain <domain>] [--generations <n>] [--resume]
select-parent
/select-parent
Select the next parent generation for mutation using configurable selection strategies. Use: /hyperagents:select-parent [--method <method>]
status
/status
Show the current state of the HyperAgents evolution process — active generation, archive health, fitness trajectory, and next steps. Use: /hyperagents:status
Agents4
ensemble-agent
/ensemble-agent
Ensemble agent that combines predictions from multiple archive generations to produce superior results. Selects the best-performing generation for each task based on archive fitness data, or uses majority voting across top-K generations.
evaluator
/evaluator
Fitness evaluation agent that scores a generation's output against domain-specific criteria. Runs the evaluation harness, generates reports, and computes fitness scores. Used by the evolve command to assess each generation.
meta-agent
/meta-agent
Self-referential meta-agent that generates code improvements. Given a target codebase and previous evaluation results, it produces diffs that improve fitness scores. Can modify any part of the code, including agent definitions, skills, and its own selection/evaluation logic. Dispatch this agent in a git worktree for sandboxed mutation.
task-agent
/task-agent
Task execution agent that solves domain-specific tasks. This is the agent that gets evolved — its code and prompts are modified by the meta-agent across generations. Evaluates tasks and returns structured predictions.
Skills6
Domain Evaluation Harness
/domain-harness
Create and configure domain-specific evaluation harnesses for the HyperAgents evolution loop. Defines how tasks are loaded, agents are invoked, predictions are collected, and scores are computed. Triggers when setting up evaluation domains or creating custom fitness functions.
Evolutionary Archive Management
/evolutionary-archive
Manage the HyperAgents evolutionary archive — an append-only log of all code generations with fitness scores, lineage tracking, and diff storage. Triggers when working with .hyperagents/ directory, archive.jsonl files, or generation metadata.
Fitness Evaluation Framework
/fitness-eval
Domain-agnostic fitness evaluation for evolved code generations. Defines evaluation harness interfaces, scoring contracts, and multi-domain aggregation. Triggers when evaluating code quality, running benchmarks, or scoring agent outputs.
Parent Selection Strategies
/parent-selection
Evolutionary parent selection algorithms for choosing which generation to mutate next. Implements random, best, score-proportional, and novelty-aware selection. Triggers when selecting parents, managing exploration/exploitation tradeoffs, or configuring evolution strategy.
Self-Referential Self-Improvement
/self-improve
Apply HyperAgents' self-referential improvement pattern to any code artifact. Triggers when Claude is asked to 'improve', 'optimize', 'evolve', or 'self-improve' code, agents, skills, or prompts. Also triggers on repeated failures as an automatic recovery strategy.
Stats
Version0.1.0
LanguageShell
Stars1
MaintenanceExcellent
LicenseCC-BY-NC-SA-4.0
Last Commit
Added
Safety Signals
Caution
Modifies files
Hook triggers on file write and edit operations
Uses power tools
Uses Bash, Write, or Edit tools
Tags
