nightly-reflection
Skill1 starsChanged 5 months agoMITView on GitHub
Read and summarize the nightly self-review logs a reflection cron writes about your assistant's recent activity. Use when the user asks what the assistant 'noticed', for recent reflections / self-reviews, or whether any patterns showed up.
In this repository3 other files it gives its agents.
- agent-backupSkill
- idea-inboxSkill
- vault-querySkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 41
Read each via your filesystem/MCP read tool and present them newest-first.
- Distinctive: line 78
Failure is silent: If the newest file is missing or empty, check the watchdog output before assuming "nothing happened" — distinguish a …
≈ “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.
Files
Nightly Reflection
A pattern for giving your assistant a memory of its own behavior. A small cron job runs each night, reads recent session activity, and writes a structured reflection to a dated file. This skill reads those files back so the user (or the assistant) can review what's been observed over time and act on it.
This is the consumer half of a two-part loop: a reflection cron writes the files; this skill reads them. The cron itself (a scheduled prompt that reviews session logs and writes
YYYY-MM-DD.md) is part of the full kit.
When to activate
- "What did the assistant notice lately?" / "recent reflections" / "self-review"
- "Any patterns this week?"
- Any question about what the assistant observed, flagged, or recommended.
How to retrieve
Step 1 — List recent files
ls -lt <REFLECTIONS_DIR>/ | head -10Files are YYYY-MM-DD.md; newest first by modification time.
Step 2 — Read the most recent N (default 5 for "recent")
Read each via your filesystem/MCP read tool and present them newest-first.
Step 3 — Summarize in your own voice, by date
For each day:
- Lead with what was observed (main activity / pattern)
- Note any friction points / "seams showing"
- Note any recommendations / "what I'd change"
A file that just says Nothing notable today. is intentional — report it briefly,
don't skip it.
Reflection file format (suggested)
# Reflection — YYYY-MM-DD
## What I observed
[the day's activity / patterns]
## Where the seams are showing
[friction, errors, thin spots]
## What I'd change
[numbered recommendations]Pairing it with a watchdog
Have a separate, tiny watchdog cron run ~30 min after the reflection cron to confirm today's file actually landed (exists + non-empty) and ping you only if it didn't. The reflection job is agentic and can fail silently; the watchdog is a dumb script, so it catches the "ran but wrote nothing" failure the reflection itself can't self-report.
Pitfalls
- Failure is silent. If the newest file is missing or empty, check the watchdog output before assuming "nothing happened" — distinguish a quiet day from a broken cron.
- Don't reinterpret quiet days.
Nothing notable today.means exactly that; don't go hunting for hidden meaning or search unrelated sources. - Keep the reader read-only. This skill consumes reflections; it shouldn't rewrite or "tidy" them — the raw log is the record.
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
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.

