← 返回 Skills 市场

Experience Distiller

Name: Experience Distiller
Author: traceme

作者 jimmyhe · GitHub ↗ · v0.1.1 · MIT-0

cross-platform ✓ 安全检测通过

164

总下载

当前安装

版本数

在 OpenClaw 中安装

/install experience-distiller

功能描述

Distill finished work into the right OpenClaw knowledge layer. Use when a task, fix, setup, integration, cron/report workflow, repeated operation, or output-...

使用说明 (SKILL.md)

Experience Distiller

Route completed work into the correct knowledge layer instead of dumping everything into one file.

Quick workflow

Read references/decision-rules.md.
Identify the finished task/result.
Separate:
- dated facts/evidence
- reusable action-level lessons
- workflow-level changes
- capability/package opportunities
Recommend one of:
- daily-log
- experience
- playbook
- skill
- multi
- no-op
If asked to execute, write the files directly.

Non-negotiables

Do not store raw noise as long-term knowledge.
Do not force everything into a skill.
Prefer experience-bank for tactical reusable lessons.
Prefer playbooks for canonical multi-step workflows.
One task may write to multiple layers when justified.

OpenClaw default mapping

memory/YYYY-MM-DD.md = dated facts and evidence
memory/experience-bank/entries/ = trigger-action-failure reusable lessons
playbooks/ = canonical workflows
skills/ = reusable capability packages

Output pattern

Use a short recommendation block:

route
confidence
why
exact files to write/update
draft content bullets

Bundled references

references/decision-rules.md — routing logic
references/template.md — lightweight invocation template
references/examples.md — ready-to-use examples for common task types

安全使用建议

This skill is internally consistent and does not request external credentials or installs. However, it can create or update knowledge files (memory/, playbooks/, skills/) if you ask it to execute — make sure you are comfortable giving the agent write access to those directories. Recommended precautions: (1) run it in a test/sandbox environment first, (2) require human review before allowing it to execute writes automatically, and (3) avoid including secrets in the task/result text you send to the skill so they don't get persisted into knowledge files.

功能分析

Type: OpenClaw Skill Name: experience-distiller Version: 0.1.1 The 'experience-distiller' skill is a knowledge management tool designed to help an AI agent categorize and store lessons learned from completed tasks into structured layers (logs, experience bank, playbooks, or new skills). The logic is entirely focused on organizational routing and documentation within the OpenClaw environment, with no evidence of data exfiltration, malicious execution, or unauthorized access in files like SKILL.md or references/decision-rules.md.

能力评估

✓ Purpose & Capability

Name/description and the bundled reference files all focus on deciding where to store post-task knowledge (daily logs, experience bank, playbooks, skills). There are no unrelated env vars, binaries, or install steps requested — the requested capabilities are proportionate to the stated purpose.

ℹ Instruction Scope

SKILL.md and references clearly define the decision logic and expected outputs. The only runtime-authorized side-effect is: 'If asked to execute, write the files directly.' That is coherent for a knowledge-routing skill, but it means the agent may create/update files under memory/, playbooks/, skills/ etc. The instructions do not instruct reading unrelated system files or exfiltrating data.

✓ Install Mechanism

Instruction-only skill with no install spec and no code files. This is low-risk: nothing is downloaded or written at install time.

✓ Credentials

No environment variables, credentials, or config paths are required. The skill does refer to repository paths (memory/, playbooks/, skills/) but does not request unrelated secrets or external service keys.

ℹ Persistence & Privilege

The skill is not marked 'always' and can be invoked by users or autonomously per platform defaults. It explicitly allows writing files when asked, which is consistent with its purpose but does grant persistence (modifying knowledge files). Confirm that the agent's file-write permissions are appropriately scoped before enabling automatic execution.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install experience-distiller
安装完成后，直接呼叫该 Skill 的名称或使用 /experience-distiller 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v0.1.1

Polish release: bundled invocation template + examples, stronger positioning, and improved post-task routing guidance.

v0.1.0

Initial release: post-task knowledge routing for OpenClaw with decision rules, template, and examples.

元数据

Slug experience-distiller

版本 0.1.1

许可证 MIT-0

累计安装 1

当前安装数 1

历史版本数 2

常见问题