corgi

Plugin corgi from marketplace corgi in Andriiklymiuk/corgi. Commands read from the repo's manifest — yesterday.

Skills (33)

• agent

  Use for corgi agent mode: Remote Control from a phone, the session board (\"who is waiting on me\"), watching trackers and PRs (\"tell me when someone comments\"), why a watch fired, and handoffs (\"leave a handoff for ABC-7\"). NOT compose authoring (corgi), starting a stack (run), debugging (debug

• align

  Use when the user says \"align\", \"let's agree first\", or a request can be read two ways before any code. Reaches a shared understanding cheaply, then codes. NOT a plan gate or a long interview.

• autopilot

  Use for a SUPERVISED loop that drains the ticket queue into draft PRs: \"autopilot\", \"keep shipping the queue\", \"drain the backlog overnight\". One spec gate per batch, never merges. NOT a single batch (/corgi-queue).

• before-after

  Use to prove a visual change against what it replaced: \"before and after\", \"screenshot comparison\", \"attach screenshots to the PR\". Captures the same screen on the base branch and yours. NOT a design match (design-parity) or store shots (mobile-screenshots).

• budget

  Use for \"how much budget do I have\", \"can I start another task\", \"when does the limit reset\". Reads `corgi agent usage --json`, answers in numbers, suggests `corgi agent carry` near a limit.

• ci

  Use to put a corgi stack in CI or run e2e across repos: \"set up CI for this workspace\", \"wire this to GitHub Actions/GitLab CI\", \"full-stack e2e on the PR\". NOT compose authoring, local runs, broken stacks or PR review.

• complexity

  Use when changed code must get simpler: \"reduce complexity\", \"this function is a jungle\", \"simplify this\", \"is this AI slop\", or as the gate stories and review run. Measures, refactors the worst first, proves behaviour unchanged. NOT performance or a whole-repo pass (quality).

• corgi

  Use to write or explain corgi-compose.yml and the corgi CLI: db drivers, healthchecks, tunnels, commands, flags. NOT starting a stack (run) or diagnosing a broken one (debug).

• debug

  Use to diagnose a corgi stack or pull runtime data for a bug: \"the api is down\", \"pull the staging logs\", \"check the 500s\", \"why is CI red\", \"works on staging not prod\". NOT compose authoring or PR review.

• design-parity

  Use to prove a build matches its design: \"does this match Figma\", \"compare it with the mockup\", \"the designer says it's off\". NOT driving a device (mobile), store shots (mobile-screenshots) or a design pass with no design (unslop).

• factory

  Use to decide how much of the work runs without a person, or to climb that ladder: \"software factory\", \"dark factory\", \"how autonomous should this be\", \"run this unattended\", \"what's our next step to more automation\", \"agents that ship without me\". NOT building one ticket (stories), one

• idea

  Use when the user wants a shareable link to a doc so others can comment: \"give me a link to the doc we made\", \"share this hld with my team\", \"дай ссылку на документ\"; to write a spec, HLD or RFC to keep; or to read and answer an idea page or /idea/<slug> link. NOT building the tickets (stories

• improve-skill

  Use to learn from a session and fix what caused the trouble: \"improve my <X> skill\", \"the skill should've caught Z\", \"make a skill for what we just did\", \"retro\", \"what went wrong in my last 30 sessions\". NOT corgi-compose (corgi) or normal code edits.

• memory

  Use to read or record the workspace memory in `.corgi/memory/`: prior decisions, incident fixes, recurring fixes worth a skill. Read before acting, confirm before writing, never secrets.

• mobile-screenshots

  Use for store screenshots of an Expo / React Native app: \"make App Store screenshots\", \"Play Market screenshots\", \"framed marketing screenshots\", \"upload screenshots to the stores\". NOT a single-screen verify (mobile).

• mobile

  Use to verify a mobile (Expo / React Native) change on a device: \"test on the emulator\", \"run it on the simulator\", \"does this screen render\", \"tap through the app\", \"drive it with Maestro\". NOT writing app code or the store submit (ship).

• prep-pr

  Use for \"prep my PR\", \"get this ready for review\", \"open a PR for this branch\". Runs the complexity gate, `corgi test --changed`, a second-reviewer pass, the risk score, then the draft PR. NOT reviewing someone else's PR (review).

• purchases

  Use for In-App Purchases, store metadata or RevenueCat in an Expo / React Native app: \"set up the IAPs\", \"configure RevenueCat\", \"sync product ids across stores\". NOT the binary submit (ship) or device checks (mobile).

• quality

  Use to raise the code quality of a repo or area: \"improve quality\", \"harden this\", \"tech debt pass\", \"audit this repo\", \"what is wrong with this code\". Baseline, real defects worst first, proof. NOT a PR review (review), one tangled function (complexity) or how a screen looks (unslop).

• resume

  Use for \"where was I\", \"pick up where I left off\", or after a session restart. Prints a six-line orientation and the next command. NOT \"what did I do today\" (standup).

• review

  Use to review EXISTING pull/merge requests: \"review this PR/MR\", \"code review <link>\", a bare PR link. Also to address feedback on your OWN: \"fix the comments on this MR\". Works from a saved PR when the forge is down. NOT creating PRs (stories) or the local diff (/code-review).

• risk

  Use to score how much human review a change deserves: \"how risky is this\", \"can this be auto-approved\", \"who should review this\", or as the gate stories, review and autopilot run. Scores 1 to 10, writes a risk card. NOT finding bugs (review).

• run

  Use to bring up a corgi-compose stack: \"run the stack\", \"run with tunnel and logs\", \"run just the api\", \"run for the android emulator\". Boots detached, waits until healthy, reports URLs. NOT authoring compose or a broken stack (debug).

• setup

  Use to set corgi up end to end: \"set up corgi for me\", \"I just installed corgi, what now\", \"connect my phone, Telegram, Jira\"; or to add a machine to an existing setup: \"set up my new laptop\", \"put corgi on the server\", \"join the cluster\", \"a second machine for the unattended work\". Ru

• ship

  Use to ship a mobile app to the stores with the repo's LOCAL build targets: \"ship it\", \"make ship\", \"build and submit to TestFlight/Play\", \"cut a build\", \"upload the binary\". NOT device checks (mobile), IAP metadata (purchases) or EAS cloud builds.

• showcase

  Use for README pictures, demos or a short video of a CLI, app or extension: \"make screenshots for the README\", \"showcase gif\", \"demo video\", \"marketplace media\". Draws the UI as HTML from the product's strings. NOT store shots (mobile-screenshots).

• standup

  Use for \"what did I do today\", \"what did corgi do while I was out\", \"write my standup\", \"what landed this week\". Runs `corgi agent today`/`standup`, reads it back, offers to post. NOT \"what's blocked\" (tracker).

• stories

  Use to ship work across a corgi workspace: a ticket key (ABC-123), a tracker link or several, \"do this story\", \"implement these tickets\"; a bare tracker link means this too. Also a free-text feature (\"add X across the services\") or \"what should I work on\". NOT authoring or running compose, o

• suggest-proactive

  Use when corgi should push ideas by itself: \"run suggest every week\", \"put the proactive bot on\", \"why did the bot propose X\", or a routine invoking /corgi-suggest-proactive. NOT on-demand ideas (suggest).

• suggest

  Use for feature or improvement ideas in a corgi workspace: \"suggest features\", \"what should we build next\", \"ideas to improve X\", \"what's missing\". NOT building it (stories), a clock (suggest-proactive) or fixing defects (quality).

• summary

  Use when someone asks where a batch of work stands: \"give me a summary\", \"what's left\", \"status of all the MRs\", \"recap\", \"which PRs are still open\". Reads live state, grouped by ticket, with links. NOT a code review or a standup post.

• tracker

  Use for tracker work in a corgi workspace (Linear or Jira): \"where are we\", \"what's blocked\", \"sort the new issues\", \"break this epic into tickets\", \"what's ready to build\". Ties tickets to real branch, PR and CI state. NOT implementing, ideas or PR review.

• unslop

  Use when a UI or a text looks AI-made, generic or too colourful: \"unslop this\", \"it looks like AI slop\", \"too many cards\", \"make it look like Linear\", \"this reads like ChatGPT\", \"make this sound human\". NOT matching a Figma (design-parity) or a one-off UI bug.

Slash commands (31)

• /corgi:corgi-align

  Agree on a request once, cheaply, then build it.

• /corgi:corgi-autopilot

  Run one iteration of the supervised autopilot loop: build-ready tickets into draft PRs, with one spec gate per batch.

• /corgi:corgi-before-after

  Capture the same screen on the base branch and on yours and attach both to the PR/MR.

• /corgi:corgi-budget

  How much budget is left: per account, the 5-hour and 7-day percent and when each window lifts.

• /corgi:corgi-ci

  Build (or fix) a CI job that boots the whole corgi stack and runs cross-repo e2e against the branches under review.

• /corgi:corgi-complexity

  Measure and reduce the cyclomatic complexity of changed code.

• /corgi:corgi-debug

  Diagnose a misbehaving corgi stack, or gather runtime data while investigating a bug.

• /corgi:corgi-describe

  Read corgi-compose.yml and produce a compact Markdown description of every service plus a Mermaid relation diagram. Writes to docs/corgi-services.md by default.

• /corgi:corgi-design-parity

  Prove a screen matches its design: pull the frames, capture the same states, compare them side by side.

• /corgi:corgi-factory

  Find how much of this workspace runs without you, and take the one next step with its brake.

• /corgi:corgi-idea

  Write and keep a page for something to build or decide, with a link people can comment on.

• /corgi:corgi-improve-skill

  Learn from this session or the last 30 and fix what caused the trouble, in a skill or a check.

• /corgi:corgi-mobile-screenshots

  Generate store screenshots for an Expo / React Native app: capture, frame, export and upload.

• /corgi:corgi-mobile

  Drive + verify a mobile (Expo / React Native) change on a device - Android emulator for JS/Skia (hot-reload), iOS simulator for native (rebuild).

• /corgi:corgi-new

  Scaffold a new corgi-compose.yml for the current project by inspecting existing service directories and asking a few targeted questions.

• /corgi:corgi-prep-pr

  Prep my PR. One entry point that runs the complexity gate, corgi test --changed against the base branch, and the risk score, then drafts the PR/MR the way stories does.

• /corgi:corgi-purchases

  Manage In-App Purchases, store metadata and RevenueCat for an Expo / React Native app.

• /corgi:corgi-quality

  Raise the code quality of a repo or an area - take a baseline, hunt real defects by kind, fix them worst first with proof, report what is left.

• /corgi:corgi-queue

  Pick up build-ready tracker tickets and hand them to the stories skill to build as draft PRs.

• /corgi:corgi-remote

  Set up or manage phone-startable Claude Code sessions on this machine.

• /corgi:corgi-resume

  Where was I. Reads corgi agent brief, corgi context, workspace memory, the session board and per-repo git state, then prints a six-line orientation and the next command.

• /corgi:corgi-review

  Code-review existing PRs/MRs against the repo standards and the linked ticket, or address feedback on your own.

• /corgi:corgi-risk

  Score how much human review a change or a story deserves, and write the risk card.

• /corgi:corgi-run

  Start a corgi-compose stack (or a slice of it) from chat - detached, then wait until healthy with a timeout and flag anything stuck.

• /corgi:corgi-ship

  Ship a mobile app to the stores with the repo's local build targets: clean prebuild, IPA/AAB, submit.

• /corgi:corgi-standup

  What did I do today: runs corgi agent standup and reads it back as done, doing, next, blocked.

• /corgi:corgi-suggest-proactive

  The Proactive bot: picks one thing worth building next and puts it on the board with its evidence.

• /corgi:corgi-suggest

  Suggest the few things worth building next in a corgi workspace, ranked, with evidence from this repo.

• /corgi:corgi-summary

  Report where a batch of work stands - 'give me a summary', 'what's left', 'status of all the MRs', 'what did we ship', 'recap'.

• /corgi:corgi-tracker

  Tracker-side planning for a corgi workspace (Linear or Jira): status, triage, or an epic split into tickets.

• /corgi:corgi-unslop

  Give an app or site a real design pass when it looks AI-generated or too colourful - neutral surfaces, one accent, rows instead of cards, measured contrast.