Skip to main content

Appendix B: Command & Skill Cheat Sheet

Trellis ships only three slash commands per platform. Everything else is an auto-trigger skill or a sub-agent. This is intentional: commands are for session boundaries; the rest of the workflow runs itself.

Slash Commands (per platform)

Trellis ships three user-invoked commands per platform (start, finish-work, continue). The invocation syntax differs per platform:
PlatformStartFinishContinueDelivery form
Claude Code / OpenCode / Cursor / CodeBuddy / Codex / Gemini CLI / Droid/trellis:start/trellis:finish-work/trellis:continueSlash command
Kiro@trellis:start@trellis:finish-work@trellis:continueSkill via @ picker
Kilo/start.md/finish-work.md/continue.mdWorkflow file (.kilocode/workflows/)
Antigravity / Windsurfworkflow picker → start / finish-work / continueWorkflow file
GitHub CopilotRun Prompt → trellis-start / trellis-finish-work / trellis-continuePrompt file (.github/prompts/)
Qoder(skill picker) start / finish-work / continueSkill — Qoder’s native Custom Commands are left free for your own use

Auto-trigger Skills

These are matched by the platform based on user intent: no explicit invocation required. Trigger them manually with your platform’s skill invocation syntax if needed.
SkillTriggers when…What it does
trellis-brainstormUser describes a feature / bug / ambiguous requestProduces task + prd.md, spawns research as needed
trellis-before-devTask is in_progress and the AI is about to write codeReads relevant spec files for the package
trellis-checkImplementation phase finishedDiff review, lint / typecheck / test, self-fix loop
trellis-update-specA learning / decision / gotcha is worth capturingAdds an entry to the right spec file
trellis-break-loopA tricky bug was just resolved5-dimension root-cause + prevention analysis

Sub-agents

Spawned by the main session via the platform’s sub-agent / Task primitive.
Sub-agentRoleRestriction
trellis-researchCodebase / doc searchRead-only
trellis-implementCodingNo git commit
trellis-checkVerify + self-fixHas its own retry loop
Available as real sub-agents on Claude Code, OpenCode, Codex, Droid. On other platforms they run inline in the main session.

task.py Subcommands

SubcommandPurposeUsage
createCreate tasktask.py create "title" [--slug name] [-a assignee] [-p priority]
init-contextInitialize JSONLtask.py init-context "$DIR" <dev_type> [--platform]
add-contextAdd context entrytask.py add-context "$DIR" <target> "<path>" "<reason>"
validateValidate JSONLtask.py validate "$DIR"
list-contextView all entriestask.py list-context "$DIR"
startSet as current tasktask.py start "$DIR"
finishClear current tasktask.py finish
set-branchSet branch nametask.py set-branch "$DIR" "feature/xxx"
set-base-branchSet PR target branchtask.py set-base-branch "$DIR" "main"
set-scopeSet scopetask.py set-scope "$DIR" "auth"
add-subtaskLink child to parenttask.py add-subtask <parent-dir> <child-dir>
remove-subtaskUnlink child from parenttask.py remove-subtask <parent-dir> <child-dir>
archiveArchive tasktask.py archive <name>
listList active taskstask.py list [--mine] [--status <s>]
list-archiveList archived taskstask.py list-archive [YYYY-MM]

Python Scripts

# Context
./.trellis/scripts/get_context.py                 # Full context
./.trellis/scripts/get_context.py --json          # JSON
./.trellis/scripts/get_context.py --mode packages # Per-package spec layers (monorepo)
./.trellis/scripts/get_context.py --mode record   # For /trellis:finish-work

# Session
./.trellis/scripts/add_session.py --title "..." --commit "..." --summary "..."

# Spec bootstrap (first-time)
./.trellis/scripts/create_bootstrap.py