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.
npx agentmods add commands/seongwoongcho/adaptive-harness/evolvegit clone --depth 1 https://github.com/SeongwoongCho/adaptive-harnessWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/seongwoongcho/adaptive-harness/evolve)<a href="https://agentmods.dev/commands/seongwoongcho/adaptive-harness/evolve"><img src="https://agentmods.dev/badge/commands/seongwoongcho/adaptive-harness/evolve.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00015 | $0.01845 |
| Opus 5 | $0.00008 | $0.00923 |
| Sonnet 5 | $0.00003 | $0.00369 |
| Haiku 4.5 | $0.00002 | $0.00185 |
Grade A, and why
evolve scanned grade A with 0 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 6d 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 200 lines — stays where its author put it; the contents beside it link to each section on GitHub.
adaptive-harness-evolve
Trigger the harness evolution cycle. Reads evaluation history, spawns the evolution-manager agent to analyze performance patterns, and applies proposed modifications to the experimental harness pool. Promoted harnesses take effect next session.
Parsing Arguments
Parse $ARGUMENTS for flags:
--skip-interview— Optional flag. If passed, auto-apply all proposals to the experimental pool without asking for user confirmation. Default behavior (without this flag) is to display each proposal and ask for confirmation before applying.
Plugin Root
Read the plugin root path: Read(".adaptive-harness/.plugin-root"). Store as {plugin_root}. All plugin-internal paths use this prefix.
Execution Steps
Step 0: Load Evolution State
Read the evolution state file to determine which sessions have already been processed:
Read(".adaptive-harness/evolution-state.json")
If the file does not exist, initialize with:
{
"evolved_sessions": [],
"evolution_memory": {}
}
evolved_sessions— list of session IDs (or eval file names) already analyzed by a previous evolution runevolution_memory— per-harness summaries from previous evolution analyses (keyed by harness name), enabling the evolution manager to build on prior insights rather than re-analyzing from scratch
Step 1: Check Prerequisites
Read evaluation history:
Glob(".adaptive-harness/sessions/*/eval-*.json")
Glob(".adaptive-harness/evaluation-logs/**/*.json")
Filter out already-processed sessions: Remove any eval files whose session ID (or filename) appears in evolved_sessions from the evolution state. Only pass NEW (unprocessed) evaluation data to the evolution manager.
If fewer than 2 NEW evaluation files found:
Not enough evaluation data to run evolution.
Current evaluations: {N} (minimum: 2)
Run more tasks via /adaptive-harness:run or auto-mode to collect evaluation data.
Step 2: Aggregate Evaluation History
Read all evaluation files. For each harness, compile:
- Total runs, success rate, average score
- Score trend (improving/declining/stable over last 10 runs)
- Most common failure modes from
improvement_suggestions - Task types where this harness underperforms
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.
- 6d ago First seen · 200 lines · 15 tokens per session scan A e2f602dc9bc1
evolve is a command published in the GitHub repository SeongwoongCho/adaptive-harness (8 stars, last pushed 5mo ago), licensed MIT. It adds 15 tokens to every session and 1,845 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
review-pr
Revisa un PR de GitHub (por URL o número, en CUALQUIER repo) con múltiples agentes en paralelo — code review por área, seguridad, consistencia de contrato, rendimiento, tests y auditoría de comentarios de revisores — verifica cada hallazgo contra el código real y guarda un consolidado. Nunca comenta en el PR; eso es…
capture-this-chat
Turn the current chat session into a reusable spell.
refine-spell
Improve an existing spell based on real usage.
create-command
Guide for creating new custom Claude commands with proper structure.
multi-mind-zen
Execute a multi-specialist collaborative analysis where every subagent delegates all substantive work to the three zen tools: mcpzenchat, mcpzenconsensus, and mcpzenanalyze.
write-llm-friendly-documentation
Orchestrates the generation of LLM-optimized documentation by spawning specialized subagents for different documentation areas. All analysis is performed on source code, not existing docs.