agent-evolver

A workflow that helps an agent learn from completed tasks by acting, checking the result, reflecting on mistakes, saving lessons, and using them later.

In plain words
What is it for?
Use it for long coding or research tasks, production changes, compliance checks, or recurring work with evaluation criteria and past lessons.
Why use it?
It reduces repeated mistakes and adds a review step for complex, high-stakes, or recurring work.

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/day1labs/openava/agent-evolver
Any agent
npx skills add Day1Labs/OpenAva --skill agent-evolver
Clone the repo
git clone --depth 1 https://github.com/Day1Labs/OpenAva

Made for: Claude Code, Codex.

Per session 44 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,038 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 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 $0.00044 $0.01038
Opus 5 $0.00022 $0.00519
Sonnet 5 $0.00009 $0.00208
Haiku 4.5 $0.00004 $0.00104

Measured 2d ago against content hash 9471d5392275, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-evolver 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 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.

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.

OpenAva/Runtime/Agent/Skills/agent-evolver/SKILL.md · 115 lines

How it starts

The opening of the file, as written. The whole thing — 115 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Evolver

Use this skill when the task is:

  • A multi-step long-horizon task (coding, data analysis, research reports) where the agent should refine strategy mid-flight and carry lessons into future runs.
  • A high-stakes task (production changes, compliance checks) that requires a built-in critic review before delivering output.
  • A recurring task where user preferences, domain SOPs, or past failure patterns should be remembered and re-applied automatically.

Inputs

  • task_goal: Clear, measurable description of what success looks like.
  • evaluation_criteria: How to judge success — automated tests, data rules, or a human review checklist.
  • context (optional): Relevant past session notes, user profile, or domain reference documents.

Core Workflow

Run each phase in sequence. Never skip Reflect or Memory Write after a task completes.

Phase 1 — Act

Plan and execute the task using the current task goal plus any lessons injected from memory.

Produce:

  • Final task output.
  • Execution log with key decisions and evidence.

Phase 2 — Evaluate

Compare actual output against evaluation_criteria.

Produce a structured evaluation with:

  • Pass/fail verdict and score.
  • Identified failure points with supporting evidence.
  • Identified success factors.

Phase 3 — Reflect

Distill the evaluation into one or more plain-text lessons — one lesson per distinct insight.

Rules:

  • Only write a lesson when there is concrete evidence (logs, test results, data snapshots). Conclusions unsupported by evidence must be tagged [confidence:low].
  • Keep each lesson to 1–3 sentences: what triggered it, what went wrong (if anything), and what to do next time.
  • Tag each lesson with short domain labels so future searches can find it.

Lesson format:

[lesson][tag1/tag2][confidence:high] When <trigger>, do <action>. Root cause if failed: <cause>.

Example:

[lesson][golang/concurrency][confidence:high] When generating Go concurrency tests, always add the -race flag and use t.Parallel() for subtests. Root cause: missing -race caused undetected race conditions.

Read the full file on GitHub · 115 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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 · 115 lines · 44 tokens per session scan A 9471d5392275

Subscribe to this mod's changes

agent-evolver is a skill published in the GitHub repository Day1Labs/OpenAva (10 stars, last pushed 1mo ago), licensed MIT. It adds 44 tokens to every session and 1,038 once invoked, about $0.0002 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.

Related

Other skills, from other repositories

pr-review-response

Teaches agents to reply to PR review comment threads after fixing issues, making resolutions traceable.

bradygaster/squad · 22 tokens

oma-explanation

Turn a code change (diff, PR, branch, commit range) into a rich, self-contained interactive HTML explainer with Background / Intuition / Code / Quiz sections. Use for explain, walkthrough, code-change explanation, diff/PR/branch explainer requests — 설명서, 해설, コード解説, 代码讲解. Produces a single offline-capable HTML file…

first-fluke/oh-my-agent · 94 tokens

creating-skills

Guide for creating Claude Code skills following Anthropic's official best practices. Use when user wants to create a new skill, build a skill, write SKILL.md, update an existing skill, or needs skill creation guidelines. Provides structure, frontmatter fields, naming conventions, and new features like dynamic context…

redai-infra/Relax · 70 tokens

llmtornado-tutorial-generator

Generates comprehensive code tutorials on LlmTornado API formatted for Medium publication with examples, explanations, and best practices.

lofcz/LLMTornado · 31 tokens

skill-creator

Generates Anthropic Skills with complete workflow including GitHub PR creation and local download verification.

lofcz/LLMTornado · 22 tokens

skill-sublation

技能扬弃 Sublation V5.0 完整版。把执行经验沉淀为观测、候选、验证、多 Agent 协作面板、默认关闭的受控唤醒意图、Skill 分类与 shadow 精准建议、用户批准、晋升和观察窗的可审计治理链路;仅在用户明确说出 sublation 或扬弃技能时触发。.

Sven-Mirana/sublation · 91 tokens