Back to Details

maintenance

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Maintenance

维护

Infra health audit. Investigate → Flag → Create issues. Never modify code, configs, or repo contents.
Install via npx:
bash
npx skills add fellowship-dev/dogfooded-skills/ops/maintenance
Architecture/doc checks have moved to 
/entropy-check
. This skill covers infrastructure only: processes, secrets, labels, branches, dependencies, system health, and memory hygiene.
基础设施健康审计流程：调查→标记→创建议题。绝不修改代码、配置或仓库内容。
通过npx安装：
bash
npx skills add fellowship-dev/dogfooded-skills/ops/maintenance
架构/文档检查已迁移至
/entropy-check
。 本技能仅覆盖基础设施相关内容： 进程、密钥、标签、分支、依赖项、系统健康状况和内存管理。

When to Use

使用场景

  • Weekly scheduled run (cron) — full infra sweep
  • After a system change (new LaunchAgent, new cron job)
  • When an automated task seems to have stopped working
  • When disk or brew issues are suspected
  • 每周定时运行（cron）——全面扫描基础设施
  • 系统变更后（新增LaunchAgent、新增cron任务）
  • 自动化任务疑似停止工作时
  • 怀疑存在磁盘或brew问题时

Guiding Principle

指导原则

Maintenance takes 10x more effort than building. This job surfaces problems Max can't see day-to-day. Output: a report + GitHub issues for anything worth tracking.
维护所需的精力是构建的10倍。本任务旨在发现Max日常无法察觉的问题。输出内容：一份报告 + 值得跟踪的问题对应的GitHub议题。

Instructions

操作步骤

Run ALL checks. Each finding either becomes a GitHub issue in fellowship-dev/claude-buddy or is noted as clean.
bash
undefined
运行所有检查项。每个发现的问题要么在fellowship-dev/claude-buddy中创建GitHub议题，要么标记为无异常。
bash
undefined

Set GH_TOKEN at the start of every run

每次运行开始时设置GH_TOKEN

export GH_TOKEN=$(grep GH_TOKEN_FELLOWSHIP /home/ubuntu/projects/fellowship-dev/claude-buddy/.env | cut -d= -f2)

---
export GH_TOKEN=$(grep GH_TOKEN_FELLOWSHIP /home/ubuntu/projects/fellowship-dev/claude-buddy/.env | cut -d= -f2)

---

1. Memory Consolidation (always)

1. 内存整合（必做）

bash
wc -l /home/ubuntu/projects/fellowship-dev/claude-buddy/memory/MEMORY.md
ls ~/.claude/projects/-Users-maxfindel-Projects-claude-buddy/*.jsonl | wc -l
  • MEMORY.md >180 lines → compact, run 
    /save-memory
  • JSONL session files: ignore accumulation — ~5MB each is fine, no action needed
bash
wc -l /home/ubuntu/projects/fellowship-dev/claude-buddy/memory/MEMORY.md
ls ~/.claude/projects/-Users-maxfindel-Projects-claude-buddy/*.jsonl | wc -l
  • MEMORY.md行数>180 → 压缩，运行
    /save-memory
  • JSONL会话文件：忽略累积情况——每个约5MB即可，无需操作

2. Secrets Scan

2. 密钥扫描

bash
grep -r "ghp_\|ghs_\|sk-\|xoxb-\|AAEi\|AAAA[a-zA-Z0-9]" \
  /home/ubuntu/projects/fellowship-dev/claude-buddy/ \
  --include="*.md" --include="*.sh" --include="*.py" --include="*.mjs" \
  -l 2>/dev/null | grep -v ".git"
Flag any file with a live-looking token/key that is git-tracked.
bash
grep -r "ghp_\|ghs_\|sk-\|xoxb-\|AAEi\|AAAA[a-zA-Z0-9]" \
  /home/ubuntu/projects/fellowship-dev/claude-buddy/ \
  --include="*.md" --include="*.sh" --include="*.py" --include="*.mjs" \
  -l 2>/dev/null | grep -v ".git"
标记任何包含可正常使用的令牌/密钥且被git追踪的文件。

3. Dependabot Coverage

3. Dependabot覆盖情况

Read 
PROJECTS.md
. For every active repo (Products + Tooling, skip FVL):
bash
GH_TOKEN=$GH_TOKEN gh api repos/{org}/{repo}/contents/.github/dependabot.yml 2>&1
  • Missing config → flag as issue
  • Config present but silent >30 days → flag
阅读
PROJECTS.md
。对于每个活跃仓库（产品+工具类，跳过FVL）：
bash
GH_TOKEN=$GH_TOKEN gh api repos/{org}/{repo}/contents/.github/dependabot.yml 2>&1
  • 缺少配置文件 → 标记为问题
  • 配置文件存在但超过30天无更新 → 标记

4. Stale Branches

4. 过时分支

For active repos, check for branches that are merged but not deleted:
bash
GH_TOKEN=$GH_TOKEN gh api repos/{org}/{repo}/branches --paginate \
  --jq '.[].name' 2>/dev/null
对于活跃仓库，检查已合并但未删除的分支：
bash
GH_TOKEN=$GH_TOKEN gh api repos/{org}/{repo}/branches --paginate \
  --jq '.[].name' 2>/dev/null

Cross-reference against merged PRs

与已合并的PR交叉验证


Flag repos with >5 stale merged branches.

---

标记存在超过5个已合并过时分支的仓库。

---

5. LaunchAgent Health

5. LaunchAgent健康状况

bash
ls ~/Library/LaunchAgents/fry.*.plist | while read f; do
  target=$(grep -o '/[^<]*' "$f" | grep -E '\.sh|\.py|\.mjs' | head -1)
  [ -f "$target" ] || echo "MISSING: $f → $target"
done
launchctl list | grep fry
Flag any plist pointing to a missing file, or any fry agent not loaded.
bash
ls ~/Library/LaunchAgents/fry.*.plist | while read f; do
  target=$(grep -o '/[^<]*' "$f" | grep -E '\.sh|\.py|\.mjs' | head -1)
  [ -f "$target" ] || echo "MISSING: $f → $target"
done
launchctl list | grep fry
标记任何指向不存在文件的plist，或任何未加载的fry agent。

6. Cron Log Errors (last 7 days)

6. Cron日志错误（最近7天）

bash
ls -t ~/.local/share/fry-bot/cron-logs/*.log | head -20 | xargs grep -l "FAILED\|ERROR\|timeout" 2>/dev/null
Recurring failures in the same job = flag as issue with log excerpt.
bash
ls -t ~/.local/share/fry-bot/cron-logs/*.log | head -20 | xargs grep -l "FAILED\|ERROR\|timeout" 2>/dev/null
同一任务反复失败 → 标记为问题并附上日志片段。

7. PROJECTS.md Accuracy

7. PROJECTS.md准确性

  • Check repos listed as active: do they still exist on GitHub?
  • Check repos listed as Dormant: any commits in last 30 days?
  • Any new repos Max created that aren't in PROJECTS.md?
bash
GH_TOKEN=$GH_TOKEN gh repo list {org} --json name,isArchived,pushedAt --limit 50
Run for: maxfindel, fellowship-dev, Lexgo-cl, CLAPES-UC, Energia-UC.
  • 检查标记为活跃的仓库：是否仍在GitHub上存在？
  • 检查标记为休眠的仓库：最近30天是否有提交？
  • Max是否创建了未列入PROJECTS.md的新仓库？
bash
GH_TOKEN=$GH_TOKEN gh repo list {org} --json name,isArchived,pushedAt --limit 50
针对以下组织/用户运行：maxfindel, fellowship-dev, Lexgo-cl, CLAPES-UC, Energia-UC。

8. Spacestation System Health

8. Spacestation系统健康状况

bash
brew outdated --quiet | wc -l
df -h / | tail -1
  • Brew packages outdated >20 → flag
  • Disk >85% used → flag
bash
brew outdated --quiet | wc -l
df -h / | tail -1
  • brew包过时数量>20 → 标记
  • 磁盘使用率>85% → 标记

9. Label Sync

9. 标签同步

Verify standard labels across active repos.
For each active fellowship-dev repo (farmesa, inbox-angel, booster-pack, mtg-lotr, commander):
bash
STANDARD="ready-to-work in-progress needs-manual-review reviewed double-checked ready-to-merge groundwork dependencies bug enhancement documentation"
for repo in farmesa inbox-angel booster-pack mtg-lotr commander; do
  existing=$(GH_TOKEN=$GH_TOKEN gh label list --repo fellowship-dev/$repo \
    --json name --jq '.[].name' 2>/dev/null | tr '\n' ' ')
  for label in $STANDARD; do
    echo "$existing" | grep -q "$label" || echo "MISSING: fellowship-dev/$repo → $label"
  done
done
  • Any missing labels → run 
    gh label create
    to add them (idempotent, safe to do inline)
  • Report how many were created vs already present
验证活跃仓库的标准标签是否一致。
针对每个fellowship-dev的活跃仓库（farmesa, inbox-angel, booster-pack, mtg-lotr, commander）：
bash
STANDARD="ready-to-work in-progress needs-manual-review reviewed double-checked ready-to-merge groundwork dependencies bug enhancement documentation"
for repo in farmesa inbox-angel booster-pack mtg-lotr commander; do
  existing=$(GH_TOKEN=$GH_TOKEN gh label list --repo fellowship-dev/$repo \
    --json name --jq '.[].name' 2>/dev/null | tr '\n' ' ')
  for label in $STANDARD; do
    echo "$existing" | grep -q "$label" || echo "MISSING: fellowship-dev/$repo → $label"
  done
done
  • 任何缺失的标签 → 运行
    gh label create
    添加（幂等操作，可直接执行）
  • 报告新增标签数量与已存在标签数量

10. Booster-Pack Sync

10. Booster-Pack同步

Pull changes into dependent sites.
Find all local repos that have a 
booster
git remote:
bash
for dir in ~/Projects/fellowship-dev/*/; do
  if git -C "$dir" remote | grep -q '^booster$' 2>/dev/null; then
    repo=$(basename "$dir")
    BEHIND=$(git -C "$dir" rev-list HEAD..booster/main --count 2>/dev/null || echo "?")
    echo "fellowship-dev/$repo: $BEHIND commits behind booster-pack"
  fi
done
  • If any site is >0 commits behind → queue a task in overnight-tasks.md: 
    pull-booster-{repo-slug}
    , type 
    maintenance
    , executor 
    {project-path}
    , prompt 
    git pull booster main --no-edit && git push
    .
  • If there are conflicts → flag for Max instead, do not pull autonomously.
将变更拉取到依赖站点。
查找所有带有
booster
git远程仓库的本地仓库：
bash
for dir in ~/Projects/fellowship-dev/*/; do
  if git -C "$dir" remote | grep -q '^booster$' 2>/dev/null; then
    repo=$(basename "$dir")
    BEHIND=$(git -C "$dir" rev-list HEAD..booster/main --count 2>/dev/null || echo "?")
    echo "fellowship-dev/$repo: $BEHIND commits behind booster-pack"
  fi
done
  • 若任何站点落后提交数>0 → 在overnight-tasks.md中添加任务：
    pull-booster-{repo-slug}
    ，类型为
    maintenance
    ，执行路径为
    {project-path}
    ，提示语为
    git pull booster main --no-edit && git push
    。
  • 若存在冲突 → 标记给Max处理，不要自主拉取。

11. Open Issues in fellowship-dev/claude-buddy

11. fellowship-dev/claude-buddy中的开放议题

List all open issues — no action, just surface in report so Max sees the backlog.
bash
GH_TOKEN=$GH_TOKEN gh issue list --repo fellowship-dev/claude-buddy --state open \
  --json number,title,labels,createdAt
列出所有开放议题——无需操作，仅在报告中展示以便Max查看积压情况。
bash
GH_TOKEN=$GH_TOKEN gh issue list --repo fellowship-dev/claude-buddy --state open \
  --json number,title,labels,createdAt

Creating Issues

创建议题

Before creating any issue, check for duplicates:
bash
undefined
创建任何议题前，先检查是否存在重复：
bash
undefined

Fetch open + closed issue titles (closed catches older attempts at the same problem)

获取开放+已关闭的议题标题（已关闭的包含之前针对同一问题的尝试）

GH_TOKEN=$GH_TOKEN gh issue list --repo fellowship-dev/claude-buddy
--state all --limit 200 --json number,title,state
--jq '.[] | "(.state) #(.number): (.title)"'

- If a matching open issue already exists → add a comment or just note it in the report. Do NOT create a duplicate.
- If a matching closed issue exists → only re-open or create a new one if the problem has clearly recurred. Note the previous issue number in the body.

For each **new** finding with no existing issue:

```bash
GH_TOKEN=$GH_TOKEN gh issue create \
  --repo fellowship-dev/claude-buddy \
  --title "Maintenance: {finding}" \
  --label "maintenance" \
  --body "..."
GH_TOKEN=$GH_TOKEN gh issue list --repo fellowship-dev/claude-buddy
--state all --limit 200 --json number,title,state
--jq '.[] | "(.state) #(.number): (.title)"'

- 若已存在匹配的开放议题 → 添加评论或仅在报告中注明。**不要创建重复议题。**
- 若存在匹配的已关闭议题 → 仅当问题明显再次出现时才重新打开或创建新议题。在议题正文中注明之前的议题编号。

针对每个**新的**无对应议题的发现：

```bash
GH_TOKEN=$GH_TOKEN gh issue create \
  --repo fellowship-dev/claude-buddy \
  --title "Maintenance: {finding}" \
  --label "maintenance" \
  --body "..."

Output

输出内容

Save report to 
reports/maintenance/YYYY-MM-DD.md
:
markdown
undefined
将报告保存至
reports/maintenance/YYYY-MM-DD.md
：
markdown
undefined

Maintenance Report — YYYY-MM-DD

维护报告 — YYYY-MM-DD

Checks Run

已运行检查项

  • Memory consolidation
  • Secrets scan
  • Dependabot coverage
  • Stale branches
  • LaunchAgent health
  • Cron log errors
  • PROJECTS.md accuracy
  • Spacestation system health
  • Label sync
  • Booster-pack sync
  • Open issues review
Note: Architecture/doc drift checks have moved to /entropy-check (separate skill)
  • 内存整合
  • 密钥扫描
  • Dependabot覆盖情况
  • 过时分支
  • LaunchAgent健康状况
  • Cron日志错误
  • PROJECTS.md准确性
  • Spacestation系统健康状况
  • 标签同步
  • Booster-Pack同步
  • 开放议题审核
注意：架构/文档偏差检查已迁移至/entropy-check（独立技能）

Findings

发现结果

Issues Created

创建的议题

  • issue fellowship-dev/claude-buddy#N: {title}
  • 议题 fellowship-dev/claude-buddy#N: {title}

Clean

无异常

  • {area}: nothing flagged
  • {领域}: 未发现问题

Noted (no issue, low severity)

已记录（无需创建议题，低严重性）

  • {finding}
  • {发现内容}

Memory

内存情况

  • MEMORY.md: {N} lines
  • JSONL sessions: {N} files

Then output a short Telegram summary:
🔧 Maintenance — YYYY-MM-DD
{N} issues created: [list with links] {N} areas clean Memory: {N}/200 lines
Full report: reports/maintenance/YYYY-MM-DD.md

---
  • MEMORY.md: {N}行
  • JSONL会话文件: {N}个

然后输出简短的Telegram摘要：
🔧 维护 — YYYY-MM-DD
已创建{N}个议题：[带链接的列表] {N}个领域无异常 内存使用：{N}/200行
完整报告：reports/maintenance/YYYY-MM-DD.md

---

What Was Removed from This Skill

本技能移除的内容

Section 11 (Imported lib drift — speckit and toolkit) was extracted to 
/entropy-check
.
Why: Checking whether speckit commands match the source of truth is an architecture/doc check — it measures whether the tooling knowledge layer is current. Entropy-check owns all doc/architecture freshness signals. This skill owns infra only.
To check speckit drift: run 
/entropy-check
against the target repo.
**第11节（导入库偏差——speckit和toolkit）**已提取至
/entropy-check
。
原因： 检查speckit命令是否与源一致属于架构/文档检查——用于衡量工具知识层是否最新。Entropy-check负责所有文档/架构新鲜度相关检测。本技能仅负责基础设施相关内容。
如需检查speckit偏差：针对目标仓库运行
/entropy-check
。