agentleFS
0xUrsanomics0xUrsanomics/utopia-os/skills/eod-summary/SKILL.md

eod-summary

Skill2 starsChanged 24 days agoMITView on GitHub

End-of-day digest pulling today's wins, decisions, status changes, follow-ups, new artifacts, tomorrow's queue from session.jsonl + tasks + tracker files. User-invoked via /eod or auto-fire at end of day.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 3 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Task-list state: all tasks added/completed/changed today (compare against yesterday's snapshot if available)

    line 21
  2. Very specific:

    When /eod fires, compile by:

    line 61
  3. Distinctive:

    Read the personal follow-up tracker for state changes

    line 65
  4. Very specific:

    progressive-disclosure: if EOD digest >1500 chars, auto-attach as .md instead of pasting

    line 87
  5. Distinctive:

    Don't speculate or pad. If a section has nothing, skip it.

    line 94
  6. Distinctive:

    Don't include yesterday's leftovers unless they materially advanced today.

    line 95
  7. Very specific:

    Timezone: always your local timezone.

    line 102

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

EOD Auto-Summary

Pull today's firehose into one structured digest. Mobile-readable wrap.

Trigger

  • /eod: manual, any time. Compiles "today so far" snapshot.
  • Auto-fire end of day. NOT WIRED. Manual only.

Source data

The compiler pulls from these locations for today (date -d "today" +%Y-%m-%d):

  1. logs/session.jsonl: events logged (commands, tasks, mode toggles, skills invoked, /save events): filter where ts starts with today's ISO date
  2. Task-list state: all tasks added/completed/changed today (compare against yesterday's snapshot if available)
  3. Git status of memory/: files modified/created today (memory writes durable across sessions)
  4. outputs/raw/ and outputs/reviewed/: files created today (work shipped)
  5. A personal follow-up tracker (e.g. workspace/personal/tracker.md): items submitted, status changes, follow-up due dates
  6. logs/errors.jsonl: errors logged today (surfaces unresolved or pending)

Format

🌙 EOD. Friday 2026-04-25

🎯 Shipped today (N):
- ...

📋 Decisions made (N):
- ...

🔄 Status changes (N):
- ...

⚠️ Open follow-ups (N):
- Due today: ...
- Due this week: ...
- Awaiting reply: ...

📝 New durable artifacts (N):
- memory/...: ...
- outputs/...: ...
- skills/...: ...

💤 Tomorrow's queue:
- ...

#meta #eod

Sections with zero items are omitted entirely (don't show empty headers).

Compilation (current: manual, model-driven)

When /eod fires, compile by:

  1. Filter today's logs/session.jsonl events (ts startswith today's ISO date)
  2. git status memory/ → new/modified files since 00:00 local
  3. List files in outputs/raw/ and outputs/reviewed/ with mtime today
  4. Read the personal follow-up tracker for state changes
  5. Pull task-list state (added/completed today)
  6. Render into the format above
  7. Save as outputs/tg/eod-YYYY-MM-DD.md, send via the reply channel

Automation (parked): an eod_digest.py script would do steps 1-7 + auto-fire at end of day. NOT BUILT. Build trigger: after manual /eod proves stable for 5+ days.

Duration / size

Target: 800-1200 char digest. If the day was huge (a big ship day + two applications + a partner brief + infra patches), the digest can stretch to 1500. Beyond that → attach as .md per progressive-disclosure.

Cadence

  • Any time during the day. Re-runs regenerate fresh from current state.
  • Soft-cap one send per day. Mid-day re-runs save to outputs/tg/ only (no spam).

Why

50-100 scattered chat messages/day → one mobile-readable artifact. Closes the loop, kills "what did I do today" decay.

Composition with other skills

  • progressive-disclosure: if EOD digest >1500 chars, auto-attach as .md instead of pasting
  • inline-tags: EOD always tagged #meta #eod for retroactive search
  • /save: EOD does NOT replace /save. /save extracts durable learnings to memory; /eod is a transient daily wrap. Both can run.
  • Stop-hook /save integration (parked): when the stop-hook runs at session end, /save fires; /eod is separate (could optionally chain at end of day, but not required).

Scope discipline

  • Don't speculate or pad. If a section has nothing, skip it.
  • Don't include yesterday's leftovers unless they materially advanced today.
  • Open follow-ups: only those genuinely actionable tomorrow. "Awaiting a contact's reply Mon/Tue" = correct. "Maybe think about a side idea sometime" = skip.
  • Decisions: include only consequential ones. "Picked B over A on a UI option" = skip. "Chose Path 1 over Path 2 for a boot-survival fix" = include.

Edge cases

  • Empty day (no shipped, no decisions, no status changes): single-line digest 🌙 EOD. {date}: quiet day. {N} open follow-ups carry to tomorrow. Skip all empty sections, no padding.
  • Timezone: always your local timezone. Use date -d "today" +%Y-%m-%d for filtering.
  • Mid-day re-run: regenerate fresh from current state, do NOT diff against prior /eod output. Each run = snapshot.
  • Partial source data (e.g. session.jsonl missing): note ⚠️ source: session.jsonl unavailable in the digest header, continue with what's readable. Don't fail the whole digest.
  • Tomorrow is weekend/OOO: 💤 Tomorrow's queue becomes 💤 Mon queue (next working day): don't pad with weekend filler unless something is genuinely due Sat/Sun.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.