daily-report

DailyReport

Low-effort daily/weekly/monthly report generator for developers. Analyzes Git activity across multiple repositories and produces Obsidian-compatible Markdown reports using AI summarization.

Prerequisites

• Python 3.11+
• Git
• Config file: ~/.config/dailyreport/config.json (created by /daily-report:dr_init or the equivalent natural-language request)

Scripts are located at: <SKILL_DIR>/scripts/ where SKILL_DIR is the directory containing this SKILL.md.

To resolve SKILL_DIR:

SKILL_DIR="$(cd "$(dirname "$0")" && pwd)" # if running from skill dir
# Or find via symlink in a home-level skills directory:
SKILL_DIR="$(dirname "$(readlink -f ~/.agents/skills/daily-report/SKILL.md)" 2>/dev/null || dirname "$(readlink ~/.agents/skills/daily-report/SKILL.md)")"

Configuration

Config path: ~/.config/dailyreport/config.json

Key fields: workspace_root, output (vault_dir, folder names), repositories (path + bound git_user), scan_settings, defaults (level, language).

See CONFIG.md for full structure and field docs.

Command Reference

Command	Purpose	Default	Params
/daily-report:dr_init	Interactive setup	Scan workspace, register repos	—
/daily-report:dr_daily	Generate daily report	standard + diff	--level=, --date=
/daily-report:dr_weekly	Generate weekly report	brief + daily summary	--level=, --diff
/daily-report:dr_monthly	Generate monthly report	brief + weekly+daily summary	--level=, --diff
/daily-report:dr_status	Show config status	Display all info	—

On tools without a slash-command menu, the same behaviors should still trigger from natural-language requests such as "generate today's daily report" or "initialize DailyReport".

See COMMANDS.md for detailed flows and examples.

Core Workflow: /daily-report:dr_init

This is the entry point. MUST be run before any report generation.

First-time initialization

1. Ask user for workspace root directory (where their Git repos live)
2. Ask user for Obsidian vault directory (where reports will be saved) Before saving config, verify the path exists. If it does not exist, ask whether to create it:

   import os
   if not os.path.isdir(vault_dir):
       # Prompt user: "该路径不存在，是否创建？(y/n)"
       # if yes: mkdir -p vault_dir
       # if no: ask again
   

   Do NOT proceed until vault_dir is confirmed to exist.
3. Ask user for default detail level (optional, defaults to standard)
4. Run repo scan:

   python3 "$SKILL_DIR/scripts/dr_scan.py" --workspace "<workspace_root>"
   

5. Parse JSON output — it's an array of discovered repos with git user info
6. Display results as a table for user confirmation:

   | # | Repo Name | Path | Git User | Email | Source |
   

   Source indicates whether git user is from local repo config or global config.
7. After user confirms, build config object and save:

   # Config is written by the agent using Write tool to ~/.config/dailyreport/config.json
   

8. Create output directories:

   mkdir -p "<vault_dir>/DailyReport/daily" "<vault_dir>/DailyReport/weekly" "<vault_dir>/DailyReport/monthly"
   

Adding new repos (re-run /daily-report:dr_init)

When config already exists:

1. Run scan with --detect-new:

   python3 "$SKILL_DIR/scripts/dr_scan.py" --workspace "<workspace_root>" --config ~/.config/dailyreport/config.json --detect-new
   

2. Show only newly discovered repos
3. Append confirmed repos to existing repositories array
4. Do NOT remove existing repos

Core Workflow: Report Generation

Shared flow for /daily-report:dr_daily, /daily-report:dr_weekly, /daily-report:dr_monthly

1. Load config — Read ~/.config/dailyreport/config.json. If missing, tell user to run /daily-report:dr_init first.

2. Determine date range and parameters:

   • /daily-report:dr_daily: date from --date param (default: today). Range = [date, date+1)
   • /daily-report:dr_weekly: current ISO week (Mon-Sun). Range = [monday, sunday+1)
   • /daily-report:dr_monthly: current month. Range = [1st, next-month-1st)

3. Determine detail level: from --level param, fall back to defaults.level in config.

4. Data collection (varies by report type — see below)

5. AI report generation: Using the collected data and report templates from TEMPLATES.md, generate the Markdown report content.

6. Write output file using Write tool to the appropriate path.

/daily-report:dr_daily data collection

Daily reports ALWAYS use diff analysis (this is the foundation of the entire system).

python3 "$SKILL_DIR/scripts/dr_analyze.py" \
  --config ~/.config/dailyreport/config.json \
  --from <date> --to <date+1> --diff

Parse the JSON output. Key fields per active repo: commits, diff_stats, diff_content, branch_activity.

If new_repos_detected is non-empty, mention it at the end of the report.

/daily-report:dr_weekly data collection

Token-efficient strategy: Weekly reports default to summarizing existing daily reports.

Decision tree:

• Check for daily report files in <vault>/DailyReport/daily/ for dates in the week range
• Has dailies + no --diff (default): Read daily report files using Read tool. Summarize them. Do NOT call Python scripts. This is the most token-efficient path.
• Has dailies + --diff: Read dailies AND run analysis:

  python3 "$SKILL_DIR/scripts/dr_analyze.py" \
    --config ~/.config/dailyreport/config.json \
    --from <monday> --to <sunday+1> --diff
  

  Use diff data to supplement/verify daily summaries.
• No dailies: Run full analysis (with or without --diff based on flag).

/daily-report:dr_monthly data collection

Token-efficient strategy: Monthly reports summarize weekly reports, which summarize dailies.

Decision tree:

• Check for weekly reports covering this month in <vault>/DailyReport/weekly/
• Check for daily reports in <vault>/DailyReport/daily/
• Has weeklies (default): Read weekly reports. Supplement gaps with dailies if needed.
• Has dailies only: Read all month's dailies and summarize.
• No reports: Run analysis. With --diff, use --stat-only flag (month-scale diff content is too large):

  python3 "$SKILL_DIR/scripts/dr_analyze.py" \
    --config ~/.config/dailyreport/config.json \
    --from <month-1st> --to <next-month-1st> --diff --stat-only
  

Three Detail Levels

brief

• Per-repo: commit count + one-line summary
• No diff details, no branch operations
• Overall: one summary sentence

standard (daily default)

• Per-repo: commit list + key file changes with descriptions
• Important diff highlights
• Branch operations if any
• Overall: summary paragraph

detailed

• Per-repo: per-commit analysis with code references
• Full diff coverage for significant changes
• Complete branch operation timeline
• Overall: detailed analysis + next steps

Report Templates

Reports use Obsidian-compatible Markdown with YAML frontmatter and tags.

Output paths:

• Daily: <vault>/<base_folder>/daily/<YYYY-MM-DD>.md
• Weekly: <vault>/<base_folder>/weekly/<YYYY>-W<WW>.md
• Monthly: <vault>/<base_folder>/monthly/<YYYY>-<MM>.md

See TEMPLATES.md for complete template structures.

Key frontmatter fields: title, date, type, level, tags, repos, generated, repos_scanned, repos_active, commits_total, authors. Reports with grouped activity should also include groups. Weekly/monthly also include: source (data source indicator), date_range.

Use > [!summary] callout for the overview section (Obsidian native support).

Script Reference

dr_scan.py

# Full scan
python3 "$SKILL_DIR/scripts/dr_scan.py" --workspace /path/to/root

# Detect new repos only
python3 "$SKILL_DIR/scripts/dr_scan.py" --workspace /path/to/root \
  --config ~/.config/dailyreport/config.json --detect-new

Output: JSON array of {path, name, git_user: {name, email}, git_user_source}.

dr_analyze.py

# Daily analysis (with diff)
python3 "$SKILL_DIR/scripts/dr_analyze.py" \
  --config ~/.config/dailyreport/config.json \
  --from 2026-04-07 --to 2026-04-08 --diff

# Weekly analysis (specific repos)
python3 "$SKILL_DIR/scripts/dr_analyze.py" \
  --config ~/.config/dailyreport/config.json \
  --from 2026-03-31 --to 2026-04-07 --diff --repos my-app,backend

# Monthly stat-only
python3 "$SKILL_DIR/scripts/dr_analyze.py" \
  --config ~/.config/dailyreport/config.json \
  --from 2026-04-01 --to 2026-05-01 --diff --stat-only

Output: JSON with date_range, active_repos[], skipped_repos[], new_repos_detected[].

Each active repo contains: name, path, git_user, commits[], diff_stats, diff_content, branch_activity.

Important Notes

1. Git safety: All operations are READ-ONLY. Never modify any Git repository.
2. Token control: Daily reports are the foundation. Weekly/monthly should prefer reading existing reports over running new analysis. The --diff flag is opt-in for weekly/monthly.
3. User binding: Each repo has its own bound git user. The --author filter uses the bound email for each repo independently.
4. Error handling: If a script fails (non-zero exit), report the error to user and suggest checking the config. If config is missing, always suggest /daily-report:dr_init or the equivalent natural-language request.
5. Language: Generate report content in the language specified by defaults.language in config (default: zh-CN, Chinese).
6. New repo detection: When analysis detects new repos in workspace, append a note at the end of the report suggesting the user run /daily-report:dr_init to register them.
7. Inactive repo presentation: In daily reports under standard and detailed, split inactive repos into:
   • recently active in the last 7 days: show repo name + last commit date
   • silent for 90+ days: list separately to surface possibly abandoned repos
   • everything else: show count only, do not expand