agentleFS
nicolas-dieznicolas-diez/personal-os/skills/calendar-colors/SKILL.md

calendar-colors

Skill4 starsChanged 9 days agoMITView on GitHub

On-demand calendar color and emoji-chip auditor. Use when asked to review or fix the colors of a week already lived or already scheduled. Reads the criteria from your own COLOR_SYSTEM.md.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write your disambiguation rules and your never-touch list (see the template).

    line 21
  2. Very specific:

    Always edit or delete: if the leading chip does not match the real category, correct it or remove the stale/duplicate one (update summary).

    line 37
  3. Distinctive:

    Never delete events: Never touch anything on the never-touch list.

    line 54
  4. Distinctive:

    Unattended runs report only: If this is ever scheduled as a background task, it reports and does not write to the calendar.

    line 55
  5. Distinctive:

    If you run a weekly planning engine, give it the same source file and run this audit as one of its steps: same criteria, one definition, no …

    line 61

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

calendar-colors

An on-demand auditor for colors + emoji chips on your calendar.

The problem it solves: color-coding a calendar is useful (you read your week at a glance, and it makes the "planned vs. real" comparison possible) but coloring each event as you create it is friction, so it never happens. This skill inverts the order: create events fast and plain, run the audit later, and the calendar ends up coded to a criterion you agreed on once.

Single source of truth: COLOR_SYSTEM.md (a filled-in copy of color-system.template.md). This skill defines no criteria of its own: it reads them from that file, and when it learns a new case, it writes it back there. Not into a parallel log.

Setup (once)

  1. Copy color-system.template.md to your private folder as COLOR_SYSTEM.md.
  2. Fill in the colorId ↔ category table. Google Calendar's colorId values are 1-11; the API returns numbers, not names, so write down which number is which color in your account.
  3. Write your disambiguation rules and your never-touch list (see the template).

Flow (every run)

  1. Load the system from COLOR_SYSTEM.md: the table, the rules, the resolved fixed cases, the emoji-chip policy. Do not work from memory: the criteria grow over time.
  2. Resolve the target week: the current week (Mon-Sun) by default, in your local timezone, checked with date rather than assumed. If a different week is specified, use that one.
  3. Fetch the events for that week from Google Calendar (list events).
  4. Classify every event against the system:
    • Correct (color + chip) → leave it.
    • No color, obvious category → fix the color automatically.
    • Wrong color where the rule is unambiguous → fix automatically.
    • Ambiguous / no definition yet → do not touch; list it for the user (step 7).
    • Never touch: anything on the never-touch list in COLOR_SYSTEM.md (other people's events, holidays, visibility placeholders).
  5. Apply the clear colors (update colorId). Report a count and one line each: event → old color → new color → the rule that decided it.
  6. Sync the title's emoji chip (conservative policy):
    • The chip is the leading emoji that mirrors the color.
    • Always edit or delete: if the leading chip does not match the real category, correct it or remove the stale/duplicate one (update summary).
    • Never add a chip to plain events that have no emoji: the color already distinguishes them.
    • Protect descriptive/semantic emoji (anything that is not one of your chips): do not touch, do not replace.
  7. Ambiguous cases → ask, one line each, with your best suggestion and the candidate rule. When the user answers, learn: append the resolved case to the "Resolved fixed cases" section of COLOR_SYSTEM.md (observable pattern → color/chip + rule + date). The learning lives in the single source of truth.

Autonomy map

LevelActions
L3, act without askingFix the color in clear cases; edit or delete an emoji chip to sync it with the color (conservative policy).
L1, ask firstAmbiguous cases or ones with no prior definition. And any event deletion. Events are never deleted (deleting an emoji is not deleting an event).
Writes to the source fileOnly into "Resolved fixed cases", and only a case the user already resolved. Never rewrite the table or the rules.

Hard rules

  • One source. COLOR_SYSTEM.md holds colors and chips. Never duplicate the table inside this skill.
  • Conservative chips. Editing or deleting a wrong chip: yes. Adding one to plain events: no. Descriptive emoji: untouchable.
  • Never delete events. Never touch anything on the never-touch list.
  • Unattended runs report only. If this is ever scheduled as a background task, it reports and does not write to the calendar. Writing happens on demand, with the user present.

Why it is worth wiring

A color-coded calendar is not decoration. It is the cheapest instrumentation you have for the week. Once every event carries a category color, "where did my week actually go" becomes a query against the calendar instead of a memory exercise, and a weekly review can diff the plan you approved against the week that happened. This skill is what keeps that data clean without adding friction to the moment you create an event.

If you run a weekly planning engine, give it the same source file and run this audit as one of its steps: same criteria, one definition, no drift.

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.