Files
awesome-copilot/skills/daily-focus-board/references/customize.md
T
jennyf19 29375678cf address review: sanitize tagc, fix customize doc, harden serve-board.ps1
- Sanitize tagc to [A-Za-z0-9_-] before class-attribute interpolation, so a custom
  class name cannot break out of the attribute and inject markup/handlers (template + sample).
- customize.md now points at .tagedit.<name> (the class the renderer actually applies).
- serve-board.ps1: fail when the port is already in use, capture and report the server PID
  (with a stop command), and verify the process did not exit before opening the URL.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 07e720ee-ca02-419e-9adb-300738b6fc76
2026-07-27 15:24:29 -07:00

1.8 KiB

Customizing the Daily Focus Board

Theming

Colors live in the :root CSS block of assets/board.template.html (--ember, --good, --urgent, backgrounds). The look is warm/dark by default. Tag colors: new (green), deadline (red), career (purple) — add your own by copying a .tagedit.<name> rule and passing that name as a task's tagc (use a plain class name: letters, digits, -, _).

v2 — file-backed state (closes the agent loop)

v1 stores progress in the browser (localStorage), which the agent can't read back. To let your AI partner read your progress (e.g. to write your end-of-day journal) and write it (e.g. "mark the design doc done" from chat), back the board with a JSON file instead:

  • Board writes state to board-state.json (via a tiny local endpoint or the File System Access API) instead of localStorage, and reads it on load.
  • The agent reads/writes that same JSON. Now it's a two-way loop: you talk, the board updates, and the agent can summarize the day from the same source of truth.

This is the genuinely agentic version — but it needs a small local write path, so it's a deliberate upgrade, not the zero-setup v1.

Optional — the "shared signals" bridge (for multi-agent workshop users)

The board's momentum notes are the same shape as an agent progress signal: a timestamped self-report of what got done. If you run a multi-agent setup that emits .signals/*.json (e.g. one signal per unit of work), you can feed those into the momentum feed so agent work and your work share one timeline — "my personal board and my agent team share one nervous system."

Keep this opt-in. The board must stand completely alone with zero workshop dependency — that's what makes it universal. The bridge is a bonus for advanced users, never a requirement.