CodeBuddy 每日工作日报生成

Generate a daily work report by automatically discovering all git repositories the user worked on, collecting commit logs across all branches, and summarizin...

Security Scan

The skill's code and instructions match its stated purpose (discover repos, collect git commits, and summarize agent sessions) but it performs a broad filesystem scan (including hidden dirs and CodeBuddy data) so review before running on sensitive machines.

Purpose & Capability

The skill claims to discover all git repos, collect commits across all branches, and gather CodeBuddy Agent session overviews — and the bundled script implements exactly that (filesystem walk, git log --all, reading a platform-specific CodeBuddy data directory). Asking to scan HOME and look for agent session files is consistent with the stated goal, though scanning the entire home (including hidden directories) is broad by design.

Instruction Scope

SKILL.md instructs the agent to run scripts/collect.py and parse its JSON output; the script walks the filesystem from HOME (and additional drives on Windows), resolves repos, runs git commands, and reads CodeBuddy's app-data directory. That scope aligns with the report goal, but it means the skill will touch many files (hidden directories, repo contents, and CodeBuddy session files) beyond just commit metadata — user should be aware it inspects the filesystem broadly.

Install Mechanism

No install spec is provided (instruction-only plus bundled script). Nothing is downloaded or executed from a remote URL by the skill itself; the README suggests an optional git clone for manual install, which is standard and not performed automatically by the skill package.

Credentials

The skill requires no secrets or external credentials. The script reads common environment variables (XDG_CONFIG_HOME, APPDATA, USER/USERNAME) only to locate directories — these are proportionate to its cross-platform discovery behavior.

Persistence & Privilege

The skill is not marked always:true and does not request permanent platform-level privileges. It reads the user's filesystem and CodeBuddy app-data but does not modify other skills or global agent configurations.

Assessment

This skill appears to do what it says: it will scan your HOME (including hidden directories) to find git repositories and will read CodeBuddy's application data to collect session overviews, then produce a report. Before installing or running it, review scripts/collect.py (you have the source), and confirm you are comfortable with a broad filesystem scan. If you have sensitive files or credentials in repos or hidden folders, consider running the script in a restricted environment (temporary account, VM, or container), or edit references/config.yaml to narrow search paths and exclude directories. Also verify there are no unexpected network calls in the code (we saw none in the provided portion) and that the generated report will be stored locally or to a location you control.

Daily Work Report Generator

Overview

This skill automatically discovers all git repositories the user committed to, collects commit history across ALL branches, gathers CodeBuddy Agent session overviews, and generates a structured daily work report. It works cross-platform (macOS, Linux, Windows) with zero configuration required. Supports generating reports for any date (today, yesterday, or a specific date).

Step-by-Step Workflow

Step 1: Run the data collection script

Execute the bundled Python script to collect all raw data:

# Today (default)
python "SKILL_DIR/scripts/collect.py"

# Yesterday
python "SKILL_DIR/scripts/collect.py" --yesterday

# Specific date
python "SKILL_DIR/scripts/collect.py" --date 2026-03-20

# N days ago
python "SKILL_DIR/scripts/collect.py" --days-ago 3

Replace SKILL_DIR with the actual path to this skill's directory (e.g., ~/.codebuddy/skills/daily-report).

Date selection rules:

• If the user says "今天" / "today" → no extra flags (default)
• If the user says "昨天" / "yesterday" → use --yesterday
• If the user says a specific date → use --date YYYY-MM-DD
• If the user says "前天" / "the day before yesterday" → use --days-ago 2
• If the user says "N天前" → use --days-ago N

The script will:

• Auto-detect the operating system
• Find common development directories
• Discover all git repos with commits on the target date (across ALL branches)
• Collect CodeBuddy Agent session overviews modified on the target date
• Output structured JSON to stdout

If the user has a custom config file at SKILL_DIR/references/config.yaml, the script will also read extra search directories and preferences from it.

Step 2: Parse the JSON output

The script outputs JSON with this structure:

{
  "date": "2026-03-25",
  "system": "Darwin",
  "git_author": "username",
  "repos": [
    {
      "path": "/path/to/repo",
      "name": "repo-name",
      "commits": [
        {
          "hash": "abc1234",
          "message": "feat: add login page",
          "branch": "feature/login",
          "time": "2026-03-25 14:30:00"
        }
      ],
      "diff_stats": "+150 -30 across 8 files"
    }
  ],
  "agent_sessions": [
    {
      "session_id": "abc123",
      "overview_content": "...",
      "modified_time": "2026-03-25 16:00:00"
    }
  ],
  "errors": []
}

Step 3: Generate the daily report

Using the collected data, generate a well-structured daily report in the user's language (default: Chinese if the user speaks Chinese, otherwise English).

Follow this structure:

Report Template

# 工作日报 - {date}

## 📊 今日概览
- 活跃仓库: {count} 个
- 总提交数: {count} 次
- 代码变更: +{additions} -{deletions}

## 🔧 项目详情

### {repo-name-1}
**分支**: {branch-names}

| 时间 | 提交说明 |
|------|---------|
| HH:MM | commit message |

**代码统计**: +{add} -{del}, {files} 个文件

### {repo-name-2}
...

## 🤖 AI 辅助工作
{Summary of Agent session overviews, grouped by project}

## 📝 今日总结
{A 2-3 sentence high-level summary of the day's work}

Step 4: Save the report

Save the generated report to the user's workspace or a location they specify. Default filename: daily-report-{YYYY-MM-DD}.md

Important Notes

• The script requires only Python 3.6+ standard library and git CLI — no pip install needed.
• If the script finds no commits, inform the user and suggest checking if git author name is correct.
• If the user wants to customize search directories, guide them to edit references/config.yaml.
• Always present the report in the user's preferred language.
• When summarizing Agent sessions, focus on what was accomplished, not implementation details.
• Group related commits across repos into coherent work items when possible.