super-evolve

A workflow for reviewing and improving an agent framework, including its memory, skills, documentation, tests, and safety checks.

In plain words
What is it for?
Use it to scan for improvements, research possible changes, run tests, focus on one area, or preview changes without applying them.
Why use it?
It helps find areas that need improvement and limits changes so framework maintenance stays controlled.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/ozmasterai/torus-framework/super-evolve
Any agent
npx skills add OZmasterAI/Torus-Framework --skill super-evolve
Clone the repo
git clone --depth 1 https://github.com/OZmasterAI/Torus-Framework

Made for: Claude Code, Codex.

Per session 35 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,833 The whole file, excluding the scripts and references it only reads on demand.
Security scan C 2 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00035 $0.03833
Opus 5 $0.00017 $0.01917
Sonnet 5 $0.00007 $0.00767
Haiku 4.5 $0.00003 $0.00383

Measured 2d ago against content hash bbff7f757b7c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

super-evolve scanned grade C with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

grep -c "gate_0[0-9]" ~/.claude/hooks/test_framework.py 2>/dev/null || \

Enumerates other installed skillsmediumAgent snooping

Other skills' SKILL.md files reveal prompts, capabilities and secrets that should be invisible to peers.

ls ~/.claude/skill-library/
dormant/skills/self-improve/super-evolve/SKILL.md · 435 lines

How it starts

The opening of the file, as written. The whole thing — 435 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/super-evolve — Ultimate Framework Self-Evolution

Merged from: /evolve + /self-improve

When to use

When the user says "super-evolve", "super evolve", "full evolution", "deep improve", or wants the most comprehensive version of autonomous framework improvement that combines component scanning, deep memory mining, external research, and structured execution.

Commands

  • /super-evolve — Full evolution cycle (scan, research, execute, validate)
  • /super-evolve --quick — Skip external research phase, memory-only analysis
  • /super-evolve --area <focus> — Limit to a specific area (gates, memory, skills, docs, tests)
  • /super-evolve --dry-run — Identify and rank improvements without implementing
  • /super-evolve --auto — Skip user approval pause (use with caution)

Hard Limits (read before executing anything)

  • Maximum 5 changes per invocation — prevents scope creep and runaway edits
  • Gates 1, 2, and 3 are protected — never modify without explicit user approval (Tier 1 safety)
  • enforcer.py, boot.py, memory_server.py — never modify without querying search_knowledge first and getting explicit approval
  • Always run the full test suite BEFORE and AFTER each change
  • Every change must be saved to memory with type:feature,area:framework tags
  • If any gate test fails post-change, revert and report — do not continue

Phase 1: SCAN — Inventory Every Framework Component

Collect the full component map. Run ALL of these:

# Gates
ls ~/.claude/hooks/gates/

# Hook scripts (registered in settings.json)
python3 -c "import json; d=json.load(open('~/.claude/settings.json')); [print(h) for h in d.get('hooks', {})]"

# Skills
ls ~/.claude/skill-library/

# MCP tools (registered servers)
python3 -c "import json; d=json.load(open('~/.claude/settings.json')); [print(s) for s in d.get('mcpServers', {}).keys()]"

# Scripts
ls ~/.claude/scripts/ 2>/dev/null || echo "(no scripts dir)"

# Plugins
ls ~/.claude/plugins/ 2>/dev/null || echo "(no plugins dir)"

Read the full file on GitHub · 435 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 435 lines · 35 tokens per session scan C bbff7f757b7c

Subscribe to this mod's changes

super-evolve is a skill published in the GitHub repository OZmasterAI/Torus-Framework (5 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 35 tokens to every session and 3,833 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it C with 2 findings (reads agent configuration directories, enumerates other installed skills). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories