WeChat
ai-engineeringai-tools
Original source

Agent Evolve — Auto-Learning AGENTS.md Maintenance for Coding Agents

Summary

Introduces “Agent Evolve” — a Harness component from Agent Plan (火山方舟) that automatically learns from past agent sessions and generates improvements to AGENTS.md/TOOLS.md instruction files. Solves the pain of manually maintaining agent config files as projects evolve.

How it works (3-step loop)

  1. Auto-review sessions — identifies repeated corrections, error patterns, reusable experience from recent task records
  2. Generate improvement suggestions — converts experience into specific diffs for AGENTS.md, TOOLS.md
  3. Apply after confirmation — human reviews diff, approves, then auto-writes back

Real-world benchmark

Same web scraping task, before vs after Evolve:

MetricBeforeAfter
Time12.7 min46.8 s
LLM calls306
Tool calls285
Token consumption4.29M321K

(Note: environment setup was already done on first run; second run benefited from reduced wrong tool calls)

5 Rules Evolve Generated

Written to AGENTS.md:

  1. Goal drift detection — every 3 steps, check alignment with original user goal
  2. Entity normalization — unify names, preserve number precision, deduplicate
  3. Structured data parsing verification — compare extracted fields with raw data
  4. Planning depth expansion — analyze preconditions, output usage, fallback plans before multi-step tasks

Written to TOOLS.md: 5. Tool parameter schema pre-validation — check types/formats before calling (dates as YYYY-MM-DD, no units in numbers, enum values only)

Setup

Supports: Claude Code, OpenClaw, TRAE

  1. Get Agent Plan API Key from 火山方舟 (Volcengine)
  2. Install Evolve Skill: curl -fsSL "https://ark-self-evolve.tos-cn-beijing.volces.com/evolve_skill/latest/install.sh" | bash
  3. Tell agent: “Set me up for evolve”

Usage:

  • “Learn from my recent sessions” → extracts patterns
  • “Show me the diff” → preview changes
  • “Apply this proposal” → write back

Docs: https://docs.volcengine.com/docs/82379/2545597