Hyper-Stigmergic Morphogenesis II

hover label filter filter

Chat

Council

Code

Context

World State

SQL

Components

Federation

Optimize

Visual

Messages

Skills

Rewards

Vault

Log

🤖 Coder Agent — Powered by pi-agent-core Ready

Ask the Coder Agent to read, write, edit, grep, find, or run bash commands.
Example: "Read src/main.rs and explain the structure"

mode: simple show rejected

Multiple LLMs reason sequentially — each challenges and builds on the previous analysis, grounded in live world data.

Ask a question to start the Socratic council...

Live data injected into the last Ollama query as grounded context

— send a chat message to see grounded context —

Loading world state...

agents beliefs improvements curious agents edge stats

Results will appear here.

Advanced Component Systems — DKS, CASS, Communication, GPU, LLM, Email

Loading component states...

Federation Management — Connect to other Hyper-Stigmergy systems

Loading federation status...

⟳ optimize_anything iteratively improve any text artifact with LLM feedback

① Artifact — what to improve

Name / label A short name so you can track runs. Doesn't affect the output.

Artifact text The optimizer will read this, score it, then produce an improved version each iteration.

examples:

② Evaluators — how to judge quality

Each evaluator scores the artifact 0–1. The optimizer uses those scores + its diagnosis to guide improvements.

③ Settings

Optimization mode

Optimize against one evaluator at a time — good for focused improvement.

Proposer model — generates improved versions

Max iterations Stops early if target score is hit.

Target score (0–1) Stop when ≥ this score.

⟳

Configure an artifact and at least one evaluator, then hit Run.

Each iteration: score → ASI diagnosis → LLM proposes improved version → repeat.

📊 Visual Explainer — Mermaid diagrams and visualizations

Use /draw, /diagram, /chart commands to generate visuals.

📊

Visual Explainer

Select a visualization from the list to view it here.
Or use chat commands like /draw architecture to create new ones.

Inter‑agent messages (last 200)

Skill evidence (recent)

Reward logs (recent)

Skill Bank — learned skills with credit scores and evolution status

JW Reward Logs — causal credit attribution from council contributions

Vault Notes — knowledge graph from markdown vault

Live event log from simulation (last 50 entries)