Files
James Montemagno e370b97107 Fix canvas repo detection and shrink oversized preview images (#2227)
Canvas extensions (accessibility-kanban, where-was-i) resolved git state
against the extension's install directory or the session-state folder
instead of the repo the session was opened in. This broke repo/branch
detection for user-scoped installs and worktrees ("unknown/unknown",
"detached HEAD"). Both now read the active session's working directory
from the canvas request context on open and on each action, so they
follow whichever project/worktree opened them.

where-was-i also re-gathers context on open when the saved branch is
blank, so it never opens stuck on stale data.

Also compress oversized extension preview images (pngquant, with a mild
resize for the photographic arcade-canvas art) so every preview.png is
comfortably under the installer's inline-byte cap that was blocking
release-notes-showcase from installing.

Update release-notes-showcase author to Kayla Cinnamon.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
2026-07-07 10:54:51 +10:00
..

Token Pac-Man

A GitHub Copilot canvas that visualizes live session AI-credit usage as a Pac-Man board. Pac-Man eats pellets as credits are consumed, ghosts chase him, fruit milestones appear, and the game ends when the configured session credit limit is exceeded.

Files

  • extension.mjs - canvas declaration, loopback server, live usage/quota syncing, and agent actions.
  • assets/preview.png - gallery preview image required by the Awesome Copilot canvas catalog.
  • assets/token-pacman.jpg - source screenshot included for the gallery.
  • copilot-extension.json - Copilot extension name/version metadata for gist installs.
  • canvas.json - Awesome Copilot gallery metadata.
  • package.json - extension metadata used by the generated website catalog.

Install

Ask Copilot to install the committed extension URL:

Install this extension: https://github.com/github/awesome-copilot/tree/main/extensions/token-pacman

The shared gist version is also available at:

https://gist.github.com/jamesmontemagno/75d701d25f49c94ba332529fb8ec1346

Agent actions

  • sync_usage - refresh the canvas from the active session's accumulated AI-credit usage and plan entitlement.
  • set_limit { limit } - set the AI-credit limit that triggers game over and resync the pellet board.
  • reset_run - clear the visible fruit streak and start a fresh chase without changing the live session credit total.