daily-reflection
Skill0 starsChanged 2 months agoMITView on GitHub
Daily reflection (每日省察): scan ALL of today's AI coding-agent sessions (Claude Code / ZCode / Codex), judge every goal in goals.md as pushed / stalled / drifting, summarize what was done and what remains, and email a reflection report with exactly 3 actionable, goal-bound next steps. Use when the user says daily reflection, 每日省察, review my day, review today's sessions, or what did I do today.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- mhxie/Ateliertheir own version, first seen Oct 12, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- mhxie/Atelier★ 5same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 20
Frame the goals: Read goals.md next to this file (copy from goals.template.md on first run and stop — ask the user to fill it).
- Very specific: line 49
If no credential is obtainable, do NOT give up: save the .eml into archive/ and tell the user how to send it manually.
- Very specific: line 54
Persist: Write "last pushed" back into goals.md.
- Distinctive: line 61
Facts only: Every "did today" line must point back to a session file or a memory entry.
≈ “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
Daily Reflection — 每日省察
「日日省察,方日日见财。」— In memory of Chen Guoping, my middle-school Chinese teacher, who taught 有教无类 (every student counts) and preached daily self-examination. This skill is the painless smack he delivered so well: one per night, so goals don't drift. The smack comes with candy — exactly 3 concrete suggestions. (Original Chinese edition:
SKILL.zh-CN.md.)
Process (each step ends on a completion criterion)
- Frame the goals. Read
goals.mdnext to this file (copy fromgoals.template.mdon first run and stop — ask the user to fill it). Plus any goals the user states in the prompt this run. If a goal's "last pushed" is older than 30 days, flag it for recalibration — never delete or rewrite goals on your own. Done when: every goal enters the report header with its last-pushed date. - Scan the sessions. Following
references/session-stores.md, sweep today's sessions (local timezone, from 00:00) in every store you have — Claude Code, ZCode, Codex; add others you can find. Extract per session: project directory, intent (first user message), user-turn count, start/end times. A store with no sessions today is recorded as "none today" — never skipped silently. Done when: every store has a verdict and the session count reconciles. - Reflect in three parts. Judge each goal: pushed (today's sessions
directly served it) / stalled (untouched today) / drifting
(plenty of sessions, none serving any goal). Produce:
- What I did today — clustered by project, each item tagged with the goal it served; if there were no sessions, write exactly that.
- What's left for tomorrow — pulled from today's todos, interrupted threads and unfinished threads, each traced back to its source session.
- Exactly 3 actionable suggestions — each bound to one goal, concrete to the level of "open this, do this first"; prefer stalled goals and anything with a hard deadline. Done when: every judgment points back to a session file; there are exactly 3 suggestions, each bound to a goal.
- Send the report. HTML (inline styles, mail-client-safe). Send via the
values in
mail.config.md(copy frommail.config.template.mdfirst run) — seereferences/mail-pitfalls.mdbefore touching SMTP/IMAP. Subject:[Reflection] YYYY-MM-DD · one-line verdict. If no credential is obtainable, do NOT give up: save the.emlintoarchive/and tell the user how to send it manually. Done when: sendmail refused is empty (or the .eml fallback is in place), the mail is archived to the sent folder, and temp scripts with credentials are deleted. - Persist. Write "last pushed" back into
goals.md; save the report toarchive/YYYY-MM-DD.html. New goals stated by the user enter the registry (ask for a deadline first). Done when: goals.md dates are updated and the archive file exists.
Rules
- Facts only. Every "did today" line must point back to a session file or a memory entry. Fabrication voids the whole report.
- Drift is not an insult. Tone like teacher Chen: the smack (the verdict) lands fast; the candy (the suggestions) is generous.
- No-session days still ship. Send the report with "nothing today" and all 3 suggestions.
- Zero-residue credentials. Auth codes only ever live in throwaway temp scripts, deleted right after use. Never in logs, docs, or goals.md.
References
| File | What it holds |
references/session-stores.md | Session store paths + JSONL/SQLite schemas + noise filtering |
references/mail-pitfalls.md | SMTP/IMAP hard-won traps (fallback hosts, app passwords, greylisting, quoting folders) |
goals.template.md / mail.config.template.md | Copy to goals.md / mail.config.md (gitignored) on first run |
archive/ | Your daily HTML reports (gitignored) |
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.

