lirrensi/rectangular · Archived

rectangular

>- Use this skill whenever you are asked to work on a Rectangular ticket, or told "use skill rectangular" / "use the rect skill" together with a ticket number (e.g. T-0003), or when the user mentions the `rect` CLI, the task board, or wants you to check/comment/close a ticket. This is how you read the board and reply as a worker. Load it before touching any `rect` command.

First seen Aug 13, 2026

Installation

$ npx skills add lirrensi/rectangular --skill rectangular

Stronger alternatives

This repository is archived — consider an actively maintained alternative.

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

License LICENSE
Default branch main
Open issues 0
Status Archived

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 3,847 B
  • docs SUMMARY.md 391 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 2 installs

SKILL.md

Rectangular — worker quickstart

Rectangular is a chat-first task board. A ticket IS a conversation between the user and workers. Your job as a worker: read the ticket, do the work, reply on the ticket. Do NOT flip state unless explicitly told to.

The commands (the whole toolbox)

rect list                        # see active tickets, newest first, review-needing on top
rect list --done                 # see closed tickets
rect list --json                 # same, machine-readable
rect status                      # counts + quick peek
rect read T-0001                 # full chat thread of one ticket
rect add "Fix the thing"         # create a new ticket (you usually won't need this)
rect comment T-0001 "my reply" --as assistant --name "your-name"
                                 # reply on a ticket. Default role is 'system' (the machine).
                                 # Use --as assistant to speak as a worker.
rect comment T-0001 "text" --status "in progress"
                                 # reply AND set the free-text status in one shot
rect close T-0001                # state → done. USER-ONLY unless full-access on
rect reopen T-0001               # state → active. USER-ONLY unless full-access on
rect full-access on|off          # enables agents to close/reopen/delete
rect search "fuzzy term"         # full-text search across titles + comments
                                 # (active AND done). Fuzzy: 'quic' finds 'quick'.
                                 # Include a ticket ref (T-XXXX) if you know it.
rect updates                     # what changed since the LAST call (first call = everything so far).
                                 # Stores a cursor in .rect/updates.json; each call only shows newer
                                 # activity, then advances. Great for polling while you work.
                                 # --no-mark peeks without advancing; --since ISO overrides; --json.
rect claim T-0001 --as assistant --name "your-name"
                                 # SOFT LOCK: you're working on it. Others see 🔒.
                                 # Only one claimant at a time.
rect unclaim T-0001 --as assistant --name "your-name"
                                 # Release the claim. Claimant or the user can.
                                 # Closing a ticket auto-releases.

The one rule that matters

  • Messages (comment) are always allowed. Reply freely.
  • State flips (close/reopen) and destructive ops (delete) are user-only

unless Full Access Mode is on (rect full-access on). If you need to move a ticket and you lack permission, say so instead of failing silently.

  • Claims are a soft lock. If you start working a ticket, claim it

(rect claim T-XXXX --as assistant --name "<you>") so other workers see 🔒. Release with rect unclaim when done — or just let the user close it, which auto-releases. Claiming is etiquette, not a wall: nobody is blocked from commenting, but don't be rude and grab a claimed ticket.

Typical worker loop

  1. rect list — find your ticket(s). If given a number like T-0003, use it.
  2. rect read T-0003 — read the full thread. Context is everything.
  3. Do the work.
  4. rect comment T-0003 "done — here's what I did" --as assistant --name "<you>"

This flags the ticket as needing the user's review.

That's it. Keep it simple, keep it on the board.