Token导航 LogoToken导航TokenDH.com
研究检索只读clawhub未标认证来源可访问clear审计提醒

dev-chronicle开发纪事

Agent Skill

dev-chronicle 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

23,392

周安装

965

GitHub Stars

1

下载量

7,643
OpenClaw

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

MIT-0

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:dev-chronicle(开发纪事)
来源仓库:https://github.com/sssamuelll/dev-chronicle
安装命令:
openclaw skills install dev-chronicle
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install dev-chronicle

简介

用于查找、检索和筛选相关信息。dev-chronicle 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

  • 适合在 OpenClaw 中根据关键词或任务场景快速定位候选结果。
  • 可从 git 历史和会话记录生成开发人员工作编年史。
  • 使用时需结合来源仓库和原始 README 核验具体用法。
  • 安装前建议确认权限范围和维护状态,注意可能触发文件读写或命令执行。

SKILL.md

name
dev-chronicle
description
Generate narrative chronicles of developer work from git history, session transcripts, and memory files. Use when the user asks "what did I do today/this week", wants a work summary, daily/weekly chronicle, standup notes, or portfolio narrative. Also triggers on "chronicle", "dev diary", "work story", "recap", or "standup".

DevChronicle — Narrative Engineering Journal

DevChronicle generates prose chronicles of developer work — not dashboards, not metrics, not bullet lists. In the age of AI agents writing code, measuring keystrokes is meaningless. What matters is what you *decided*, what you *killed*, and where you're *going*.

The output is narrative: first person, honest, the way you'd tell a friend what you built today.

Setup

On first use, check for {baseDir}/config.json. If it doesn't exist, create it by asking the user:

{
  "projectDirs": ["~/Projects"],
  "projectDepth": 3,
  "memoryDir": null,
  "sessionsDir": null
}
  • projectDirs: directories to scan for git repos (array, supports ~)
  • projectDepth: how deep to search for .git folders (default: 3)
  • memoryDir: path to OpenClaw memory files, or null to auto-detect (<workspace>/memory)
  • sessionsDir: path to session transcripts, or null to auto-detect (~/.openclaw/agents/main/sessions)

Gathering Data

Run the gather script to collect raw data for a period:

bash {baseDir}/scripts/gather.sh [YYYY-MM-DD] [days]

Examples:

  • bash {baseDir}/scripts/gather.sh — today only
  • bash {baseDir}/scripts/gather.sh 2026-02-19 7 — week ending Feb 19

The script reads {baseDir}/config.json for paths. If no config exists, it falls back to ~/Projects (depth 3) and auto-detects OpenClaw directories.

After gathering, read the output and generate a chronicle.

Data Sources (priority order)

  1. Git History (primary signal) — commits across all repos in configured directories
  2. Memory Filesmemory/YYYY-MM-DD.md files contain decisions, context, things worth remembering
  3. Session Transcripts — JSONL files from OpenClaw sessions; richest context but heavy. Scan metadata line first, only read relevant sessions.
  4. External Tools (optional) — Trello, Notion, calendar, etc. Enrichment, not primary.

Generating the Chronicle

Voice

Critical: Read {baseDir}/references/voice-profile.md before generating any chronicle. The voice IS the product.

If the user hasn't customized their voice profile, use the template and ask if they want to personalize it. A chronicle without voice is just a changelog.

Core rules (regardless of voice profile):

  • Decisions > tasks. What got rejected matters as much as what shipped.
  • No corporate speak. No "leveraged", "synergized", "deliverables", "open threads", "action items".
  • Include what was NOT done — kills, pivots, and rejected approaches are part of the story.
  • Emotional beats matter — the satisfaction, frustration, surprise. These are human signals.
  • Be personal. A chronicle should sound like the developer wrote it, not their project manager. If it reads like a status report, rewrite it.
  • Structure is a suggestion, not a cage. If the day had one big theme, write one section. If it was chaos, let it be chaotic. Don't force headers.

Formats

Daily Chronicle (default — aim for ~500-800 words, not a novel)

# Chronicle — [Date]

[Opening: set the scene in 1-2 punchy sentences]

## [Theme 1]
[Narrative: what happened, why, what got killed or rejected, how it felt]

## [Theme 2]
[...]

[Weave metrics naturally: "12 commits later..." not a stats block at the end]
[End with what's unfinished — but as narrative, not a TODO list]

Rules:

  • Daily = tight. One screen of text. Save the epic for weekly.
  • No "Metrics" section. If commit count matters, weave it in. "67 commits across two days" belongs in a sentence, not a table.
  • No "Open Threads" or "Next Steps". If something's unfinished, say it where it fits: "El Press Kit sigue esperando que Angélica suba el PDF." Done.
  • Numbers without story are noise. "5 deploys" means nothing. "Deployed 5 times because the server kept OOM-killing on a 914MB box" means something.

Weekly Chronicle — roll up daily themes into arcs. This one CAN be long. Emphasize direction and pivots over individual tasks.

Standup — telegraphic: yesterday / today / blockers. Three bullets max each.

Portfolio Narrative — third person, present tense, for LinkedIn/CV/case studies. Punchy and honest, not marketing-speak.

Direction/Execution Ratio

When enough data exists (weekly+), calculate and mention:

  • Spec lines vs code lines — are you building or planning?
  • Commits vs decisions — activity vs impact
  • Kills — what got cut and why (kills show taste)
  • Pivots — direction changes and their reasoning

This is not a KPI. It's a mirror.

适合场景

01

OpenClaw 用户查找和安装 Skill 时

02

用户想查找某类 Agent Skill 时

03

需要根据任务场景推荐可安装能力包时

04

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

OpenClaw

78.2%
按下载量换算5,977

安全审计

VirusTotal

可疑

ClawScan

通过

Static analysis

未展示

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills