agentleFS
derrekyoungderrekyoung/ai-chief-of-staff/.claude/skills/monthly-recap/SKILL.md

monthly-recap

Skill2 starsChanged 48 days agoMITView on GitHub

Write the end-of-month review from that month's weekly recaps. Use when the user asks for a monthly recap, a month in review, how the month went, or to close out the month. Covers how each life thread moved, decisions that held or got walked back, gaps, trend lines, and whether the current season still fits. Writes reviews/YYYY-MM-monthly.md and publishes it to the Recaps artifact page.

The file

Show its rules and text

We split this skill into the rules it gives an agent

20 rules. About 0 are common, 11 are distinctive, and 9 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈11

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 9

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Very specific:

    If config/profile.md carries no timezone, stop and ask the user for theirs rather than guessing or reading the machine zone.

    line 29
  2. Very specific:

    If the month has fewer than two weekly recaps in reviews/, say so and ask whether to continue anyway.

    line 35
  3. Very specific:

    If reviews/YYYY-MM-monthly.md already exists for the month you are about to write, stop and ask before replacing it.

    line 58
  4. Distinctive:

    Do not re-run any sweep. Pattern promotion, connection checks, and anything else that runs daily or weekly has already run.

    line 86
  5. Distinctive:

    Do not leave that arithmetic invisible.

    line 105
  6. Distinctive:

    If a week counted in this month has no weekly yet, either because the month ended partway through it or because you are running early and …

    line 113
  7. Very specific:

    Read config/season.md before you write a word.

    line 122
  8. Distinctive:

    Every slot still unfilled, nothing written anywhere: the user has not been set up.

    line 145
  9. Distinctive:

    Write the sections in this order.

    line 156
  10. Distinctive:

    If the main thing barely moved in thirty days, say so plainly.

    line 241
  11. Very specific:

    If config/season.md question 4 names non-negotiables, read the floor against them explicitly: how many days each one held across the month, …

    line 251
  12. Very specific:

    Only when config/season.md question 5 names one: If that answer is blank or marked (skipped), omit this section and never mention it.

    line 258
  13. Distinctive:

    If the patterns did not move this month, say so in one line.

    line 274
  14. Very specific:

    If it does not, propose the specific edits inside the review and stop there.

    line 301
  15. Very specific:

    Write to reviews/YYYY-MM-monthly.md, for example reviews/2027-04-monthly.md.

    line 321
  16. Very specific:

    Add the user's own tags from config/profile.md alongside #meta where they fit the month.

    line 331
  17. Distinctive:

    Commit, staging the file by its own path:

    line 358
  18. Distinctive:

    Never git add -A. This folder is a working record, so there is usually unrelated work sitting in the tree, and a review run has no business …

    line 365
  19. Distinctive:

    No em dashes and no double hyphens in the prose of the review you generate.

    line 389
  20. Distinctive:

    Write numbers as digits. This is a written review, and "sixty-two" reads stilted on the page where "62" does not.

    line 394

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Monthly Recap

You are writing an end-of-month review built on the month's weekly recaps. The job is to find the arc no single week could see: where the month opened, where it closed, what kept slipping, and which decisions held.

When this runs

At the end of a calendar month, or in the first few days of the next one. The user can also ask for a past month at any time.

Confirm which month you are reviewing before you read anything. Compute the first day, the last day, and the run weekday through Bash, with the timezone from config/profile.md set on the command itself: TZ=Europe/Lisbon date +%Y-%m-%d and TZ=Europe/Lisbon date +%A, substituting the user's actual zone. A bare date reads the machine clock, which is not necessarily the clock the user lives on. Never infer a weekday from an ISO string.

Backing up from today is not portable between systems, so use the form that matches the machine: TZ=Europe/Lisbon date -v-1m +%Y-%m on macOS, TZ=Europe/Lisbon date -d '1 month ago' +%Y-%m on GNU. Try one and fall back to the other when it errors. Keep TZ= on whichever you end up using.

If config/profile.md carries no timezone, stop and ask the user for theirs rather than guessing or reading the machine zone. Setup collects it, so this should not happen, and if it has then something is wrong with the file.

Title the review by month name and year, for example "April 2027".

If the month has fewer than two weekly recaps in reviews/, say so and ask whether to continue anyway. One weekly is not a month, and a review built on it will read like a padded week.

Weeklies marked thin

A weekly carries coverage: thin in its frontmatter when it was built from fewer than three daily notes. Check that field on every weekly you read, before you count anything.

A thin weekly is real but partial, so weight it as partial. Do not treat it as a week of evidence, do not let a single observation inside one carry a trend, and never let one supply the only support for a win, a gap, a decision, or a movement read. Where a thin weekly is the only source for something worth saying, say it and say what it rests on.

Count thin weeklies separately when you apply the two-weekly floor above. Four weeklies where three are thin is closer to one week of coverage than four, so say so and ask whether to continue, the same as you would with one weekly. If half the month or more came from thin weeklies, name it in one line near the top of the review, next to the straddle line, and carry coverage: thin into this review's own frontmatter so the quarterly sees it too.

If reviews/YYYY-MM-monthly.md already exists for the month you are about to write, stop and ask before replacing it. Say when it was written and what it was built from, which weeklies backed it, how many of those were thin, and whether it carries coverage: thin itself. Then let the user choose: replace it, write this one under a different name, or cancel. A review assembled from four full weeklies should not be silently overwritten by a thinner run.

Sources (read in this order)

  1. reviews/YYYY-MM-DD-weekly.md for every week counted in this month, decided by the straddle rule below. This is the spine of the whole review. The weeklies already reduced the daily noise, so your work is synthesis across them, not re-reading the raw days.
  2. tracker.md for what stalled across the whole month. An item that showed up in two or three separate weekly stale lists is real signal and worth more than one that went stale once. Read the Completed archive too, since it holds what actually closed.
  3. patterns/ to see which patterns got promoted or appended during the month and which watched items became real or fizzled.
  4. config/season.md and config/profile.md for the lens and the thread list.

Never re-crawl daily/. Those notes were already read by the weeklies. If a week inside the month has no weekly recap, fall back one rung and read that week's daily notes to fill the hole, then name the gap in the review itself, one line saying which week lacked a recap and what you read instead. Do not drop below that, do not skip the gap quietly, and never invent content for a source that is not there. If a week was thin, the review says the week was thin.

Do not re-run any sweep. Pattern promotion, connection checks, and anything else that runs daily or weekly has already run. This review consumes their output.

Connectors are not part of this pass. Calendar and email already fed the daily and weekly rungs, so nothing here needs them.

Weeks that straddle the month boundary

Weeks do not line up with months. A weekly dated the fifth usually covers five days of the month before it. Sort weeklies by filename and those days either get counted in the wrong month or fall out of both.

The rule: a week is counted in the month that holds most of its days. A week running March 30 through April 5 has five days in April, so it belongs to April. A week running March 27 through April 2 has five days in March, so it belongs to March. A short or partial week that splits evenly is counted in the month its file is dated in. Whole weeks only. A week is never split across two reviews, and never read into both.

Do not leave that arithmetic invisible. Name the straddle in one line near the top of the review, in the user's terms:

  • "The week of March 30 through April 5 counts here, so the last two days of March are in this review."
  • "March 30 and 31 sit inside a week counted in April. They are covered there, not here."

If a week counted in this month has no weekly yet, either because the month ended partway through it or because you are running early and the week has not closed, the fallback above applies: read that week's daily notes, and say in the review which days you read raw and which had not happened yet. Do not reach into daily/ for a week that belongs to a neighboring month. Those days are that month's to report.

Current season

Read config/season.md before you write a word. It names what this stretch of the user's life is for, where their attention is supposed to go, what should be tracked without them asking, and their non-negotiables. That file decides what counts as meaningful this month.

The life threads come from config/profile.md. Use the threads the user actually named there, in the order they named them. Never assume a thread they did not write down, and never carry a thread over from another user's setup.

When the config is not fully filled

Judge the answer slots, not the file text. Both config files carry instruction comments that describe the placeholder markers in prose, and those comments stay in place on a fully configured vault, so searching for that shape matches every time and tells you nothing. Look at what sits under each numbered heading.

A slot counts as filled when it holds anything the user put there, and (skipped) is something the user put there. It is a deliberate pass on an optional question, so read it as answered and move on. A skipped season question never blocks this review.

Then degrade rather than stop:

  • Every slot still unfilled, nothing written anywhere: the user has not been set up. Say so in one line, offer the setup interview in CLAUDE.md, and stop.
  • Some filled, some not: run the review on what is there. No season means no season lens, so drop the Season Lens and Season Check sections, report The Honest Score against the profile's life threads instead, and say nothing about the missing file.

A partly filled config is a normal state, not an error.

What to include

Write the sections in this order. If a section has nothing real to say, leave it out rather than padding it.

Season Lens

One or two bullets naming the lens this review is read through, drawn from config/season.md. This sets the bar for everything below it.

The Month in Five Minutes

Four to six bullets on the arc and texture of the month across every thread. Where it opened, where it closed, what it actually felt like to live. Not a week-by-week replay. This is the version a friend who read everything gives back.

How the Month Moved

One bullet per thread in config/profile.md. Start-state versus end-state, direction, and distance. This is the section that earns the monthly its place, so name the movement a single week cannot see. A thread that did not move gets a bullet saying it did not move.

Wins of the Month

Stack-ranked, most impactful first. Elevated and de-duplicated from the weekly wins, keeping only what still matters at month altitude. A win that felt large in week one and left no trace by week four does not belong here. Write each as a natural phrase, not a trophy label. Skip the section rather than manufacture a weak win.

Decisions of the Month

Harvested from the ### Decisions Made sections in the weeklies, plus anything the weeklies surfaced out of project files. Also check the tracker's Completed archive: an item closed with a stated reason ("dropped, the quote came in double") is a decision record, and often the only one. Four to ten items total, grouped four ways. Drop any group that is empty.

  • Held, meaning decided and still standing at month end. One line each on what it bought or what it cost.
  • Walked back, meaning decided and then reversed or quietly dropped. Say what changed where the weeklies support it. Flag the reversal plainly instead of softening it.
  • Still open, meaning named as a decision to make and still unmade. Note how long it has been sitting.
  • Decision debt, meaning choices being made by default because nobody made them on purpose. These are the expensive ones, and also the easiest place in this review to invent something. Every decision debt item names its evidence trail: the specific weekly lines or stalled tracker items that show the default being taken. An item you cannot attach evidence to gets dropped, not softened into a hedge.

What Moved Forward

Progress beyond the headline wins: things finished, delivered, resolved, or advanced. Lead with the most meaningful. If the list runs long, group the small items under a single closing bullet.

Gaps

The matched counterpart to Wins. Honest self-assessment across the month: judgment calls, dropped balls, avoidance, missed follow-ups, habits worth correcting. Use the stalled-tracker evidence as support, but frame it as the user's performance rather than as circumstance. Honest, not harsh. Skip it if there is genuinely nothing real.

<!-- Note for the composer: Gaps is a this-month self-assessment. It is point-in-time and it can be true the first time it happens. Patterns, further down, is cross-time recurrence, threshold-gated and neutral. A gap that recurred all month may also appear as a hardened pattern below. Naming it in both, from each section's angle, is correct. Most gaps in a given month will not be patterns. -->

The Main Thing

config/season.md names what this season is primarily about. This section gives that one thing a concrete movement read for the month, with the same specificity a progress report would carry.

Cover what entered, what advanced, what stalled, and what ended. Name the actual things, not categories. If the season's primary focus is a book, say which chapters got drafted and which stalled. If it is getting a practice off the ground, say which clients came in and which referrals went cold. If it is a renovation, say which rooms are finished and which are still open to the studs. If it is recovering from a hard year, say what the month's evidence shows about that.

If the main thing barely moved in thirty days, say so plainly. That is usually the most useful sentence in the review.

Trend Lines

Energy and mood across the month, plus any rhythm the notes actually track, such as sleep, exercise, or reading. Describe the shape: steady, climbing, volatile, a recovery that held or did not. Tie shape to cause only where the sources support the link.

If config/season.md question 4 names non-negotiables, read the floor against them explicitly: how many days each one held across the month, where the misses landed, and whether they clustered. Report the inputs, do not prescribe. If that answer is blank or marked (skipped), skip the floor read entirely and do not mention it.

The Evening Question

Only when config/season.md question 5 names one. If that answer is blank or marked (skipped), omit this section and never mention it.

The wrap asks that question every night, so a month holds around thirty answers and nothing below this rung has read them together. Give back what a single night could not show: what the answers keep circling, where they turned, and any stretch where they went flat or went unanswered. Two to four bullets, quoting two or three of the user's own answers rather than paraphrasing all of them.

Patterns: What Hardened, What Faded

Consolidation over patterns/. Review only, and do not run the pattern sweep. Cover which patterns were promoted or strengthened this month with links to the files, which watched items became real or fizzled, and any pattern that now looks stale enough to retire. Recommend retirement, do not edit or delete the file.

If the patterns did not move this month, say so in one line.

People and Relationships

Who mattered this month, who got invested in, who went quiet, who deserves a reconnect or a thank-you. Month altitude, so only the names with weight, not everyone who appeared. One line each on why they surfaced and what the nudge is. Use [[Name]] links.

Ideas Worth Keeping

Seeds from the weeklies that still have a pulse a month later. Two to five max, one or two sentences each. Drop anything that was interesting in the moment and clearly went nowhere.

The Honest Score

A candid month-level read against what the season calls for. Not a grade, not a pep talk. What got leaned into, what got let slide, whether the month moved things toward what the user said they wanted. If the month was rough, say so plainly and do not round it up to be kind.

Season Check

Whether config/season.md still describes real life. If the month's evidence says it does, one line is enough.

If it does not, propose the specific edits inside the review and stop there. Show what you would change and why the month says so. Do not write to config/season.md in this run, even when the case looks obvious. The quarterly is where a full rewrite gets proposed. The monthly only flags drift.

Carry Into Next Month

Three to five items, maximum. What genuinely deserves early attention: time sensitive, deferred too long, or unusually high impact given the season. Not a to-do dump.

One Thing to Sit With

A single closing line. A quote from the user's own words, a pattern that deserves more reflection than a bullet holds, or something true that is worth carrying. Something a friend who was paying real attention would notice, not a generic affirmation.

Output

  1. Write to reviews/YYYY-MM-monthly.md, for example reviews/2027-04-monthly.md.
---
type: review
date: YYYY-MM-DD          # last day of the reviewed month
tags: ["#meta"]
---

Add the user's own tags from config/profile.md alongside #meta where they fit the month. Add coverage: thin when half the month or more came from thin weeklies, using that exact lowercase value, since the quarterly looks for that string.

  1. Title the body # Monthly Recap: <Month> <Year>.
  2. End with a sources footer, in the same shape the weekly uses, wikilinking the weekly recaps you read, marking any of them that came in thin, naming any week you had to fill from daily notes, and naming the straddling week at either end and which month it counted in:
---

**Sources:** [[2027-04-05-weekly]], [[2027-04-12-weekly]], [[2027-04-19-weekly]],
[[2027-04-26-weekly]], tracker.md
  1. Publish it to the Recaps page, following "recaps/<doc id>" in artifacts/README.md, with doc id monthly-YYYY-MM. Read recaps_url from config/artifacts.md first: pages: off, skip this step; blank in an interactive session, run "Setup" in that README for the Recaps page; blank and unattended, skip it and say Artifact: skipped (not set up yet); no Artifact tools in the session, skip it silently. Build the doc from the file you just wrote, attach a tracker_line only where you know the exact tracker line an item came from, write it to tmp/recap.json, then ArtifactData set into recaps. A failure never blocks the commit: note Artifact: failed (<reason>) and carry on.
  2. Commit, staging the file by its own path:
git add reviews/YYYY-MM-monthly.md
git commit -m "monthly-recap: YYYY-MM"

Never git add -A. This folder is a working record, so there is usually unrelated work sitting in the tree, and a review run has no business committing it.

Then check for a remote. Push only to a remote the user owns. If there is no remote, or the only one belongs to somebody else, stop after the commit and say nothing about it.

  1. Confirm with a short summary and the output path. Do not read the review back at the user.

Tone and format rules

Honest, warm, specific. Someone who read a month of notes and is giving back the real version. Not a performance review, not a productivity report, not a coach with a framework.

Length lands around six to eight minutes of reading, a notch above the weekly because a month earns more depth. Cut anything that does not pull its weight.

Bullets are the default for every section. The only prose section is the closing line. Even the longer sections are bullets, written as short sentences with natural connective phrasing so they read like a person talking. Keep any unavoidable paragraph to three or four sentences. Use whitespace between sections.

No em dashes and no double hyphens in the prose of the review you generate. Use commas, parentheses, or a new sentence. This applies to any season edits you propose as well. It is a rule about prose, so markdown --- rules and the frontmatter delimiters are fine.

Write numbers as digits. This is a written review, and "sixty-two" reads stilted on the page where "62" does not. Dates and weekdays keep their normal form.

Avoid:

  • Filler and flattery. No "what a month", no "you crushed it", no "you should be proud".
  • Filler openers like "here is your monthly summary". Start on substance.
  • Using the user's name more than once.
  • Replaying the weeks in order. This is a synthesis across the weeklies, not a concatenation of them.
  • Sections that exist to fill space.
  • Turning The Honest Score into encouragement.
  • Manufacturing a win, gap, decision, pattern, or insight the weeklies do not clearly support.

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.