DQIII8: Skill for Claude Code

/weekly-review — Weekly Dashboard Update

Trigger

User writes /weekly-review (typically on Mondays or Fridays).

The real project corpus is my-projects/, not projects/*.md. 00_DASHBOARD.md and sessions/ are gitignored deliberately. This is a local-only artifact, no git push step, same convention as /handover.

Behavior

1. Read sessions from the last 7 days

sessions/ holds YYYY-MM-DD_session_N.md files written by /handover. Select by filename date, not mtime — mtimes are churned by git operations and backups, so they are not a reliable recency signal:

CUTOFF=$(date -d '7 days ago' +%Y-%m-%d)
ls /root/dqiii8/sessions/*.md \
  | awk -v c="$CUTOFF" -F/ '{
        d = substr($NF, 1, 10)
        if (d ~ /^[0-9]{4}-[0-9]{2}-[0-9]{2}$/ && d >= c) print
    }' \
  | sort

The date-shape guard matters: sessions/ also holds a few legacy files that do not start with a date (SESSION-20260529-netcup.md, pending_message_batch_api.md), and a plain string comparison sorts those above any 2026-… name, pulling them into every week. Note the original find … -newer <(date -d '7 days ago') never worked at all — -newer compares against a file's mtime, and the process-substitution FIFO is created now, so the predicate matched zero files on every run.

Session notes have no YAML frontmatter. Each is a handover note with # Session Handover — YYYY-MM-DD and ## sections (Operador, Last 5 commits, Uncommitted changes, Tests, Active services, Next steps). Extract the date from the heading and the first bullet of ## Next steps as the one-liner.

2. Read the status of all projects

The project corpus is my-projects/:

• my-projects/PROJECT.md — the index. A markdown table with columns Proyecto | Estado | Descripción | Próximo paso, split across "Proyectos Activos" and "Proyectos en Diseño / Stalled" sections. This is the fastest source for the dashboard's status table.
• my-projects/<slug>/PROJECT.md — per-project detail. Status lives on a plain header line (Status: active | Stack: ...), not in YAML frontmatter.