factory

Plugin factory from marketplace factory in MasonGeloso/factory. Commands read from the repo's manifest — 8 days ago.

Skills (46)

• factory-ads-assess

  The recurring ads review — usually weekly, with the owner. Pulls each campaign's spend and clicks from the ad platform and its signups from the project's own attribution, separates measured from hand-set and presumed conversions, flags confounds, compares against the baseline, and ends in ONE recomm

• factory-ads-create

  Build one ad, end to end — pick an angle, verify the feature it sells by actually opening it, write copy as plain declarative sentences, brief a graphic to a design-capable model, render both languages, self-check, and hand it to the `factory-ads-qa` gate before any money is pointed at it. Reads per

• factory-ads-onboard

  Interactive, hands-on onboarding for the ads pillar — sets up `factory/ads/platforms.md`, `budget.md`, `attribution.md`, `offers.md`, and the `strategies/`, `visuals/`, `ads/` directories. Like the marketing onboarding and unlike the rest of Factory's, this is not a batch of upfront questions — afte

• factory-ads-post

  Launch one finished ad on an ad platform, with the user's explicit go-ahead — set the campaign, the budget caps, the schedule and the targeting, upload the creative, walk to the review screen, and publish. Spends real money, so it never runs on its own initiative and never guesses a value that isn't

• factory-ads-qa

  The QA gate for ad creative — multi-round, external, fresh-eyes review of a finished ad before any money is spent on it. Reviews the rendered image and its copy against generic advertising discipline (legibility at feed size, one clear subject, claim integrity, brand consistency, layout paper-cuts)

• factory-ai-pm

  Run as an automated AI Product Manager. On each (usually scheduled) run, read the org's communication channels since the last check, reconcile the work tracker against the conversation — create, close, assign, and comment on issues — then save a new last-read timestamp. Reads per-project rules from

• factory-code-review

  Independent, fresh-eyes review of one pull/merge request — reads the full diff, the linked issue, and the project's own checklist (`factory/code-review.md`), scores generic engineering categories (correctness, security, duplication/simplification, conventions, test coverage, docs), then posts a PASS

• factory-communication-setup

  Onboard how an org communicates so Factory agents (the AI PM and future ones) can read from and — where permitted — write to its channels. Interviews the user about the medium (Slack, Discord, …), authentication, which channels to read, and reply policy, then writes `factory/communication.md`. Use w

• factory-content-remember

  Given one finished piece of marketing content and feedback about what's wrong with it, track down the session that produced it, pull the actual published/drafted text, diagnose which step of the pipeline let the defect through, and propose (never auto-write) a checkable rule for `factory/content-rev

• factory-content-review

  Independent, fresh-eyes review of one finished piece of published-facing content (a post, article, card, reply, recap) — the content-side counterpart to `factory-code-review`. Reads the finished text, the full source-material packet behind every checkable claim, and the project's own config (`factor

• factory-daily-digest

  Produce a concise daily digest of what's happening across an org's systems. Surveys recent activity (commits/diffs, tracker changes, conversation) over the last day or two, writes a short bullet-point TLDR of what's being worked on, and posts it where the org wants — factoring in current priorities

• factory-demo-light

  Minimum-sufficient proof that a small/medium change works — one screenshot, a short clip only if something animates, a log excerpt for backend/pipeline work, or a Claude Artifact link for something visual/interactive. Not the full demo package (`factory-demo-video`/`factory-demo-terminal` produce a

• factory-demo-terminal

  Factory-suite copy of the demo-terminal skill. Record polished terminal-style screencast demos by capturing real CLI output and replaying it in an animated HTML "terminal" page via Playwright. Outputs a .webm plus a numbered PNG still of every scene and an artifacts.md manifest, so the work can be r

• factory-demo-video

  Factory-suite copy of the demo-video skill. Create polished demo videos of web applications using playwright-cli with HTML overlay cards, code diffs, chapters, and banners. Records .webm video of a real browser session with composited overlays, and also captures numbered PNG screenshots of every key

• factory-design-iso

  Build an interactive isometric line-art "figure" (Fig N style) of an object in a single HTML/SVG file: monochrome technical-illustration look, parts you can press that produce real output, live state readout. Use for hero illustrations, interactive diagrams, product figures, and playable desk object

• factory-execute

  The /execute (build) step of the factory pipeline — take an ALREADY-FORMED, heavily-researched plan and build it: set a granular todo list and implement it one item at a time, verifying as you go, never stopping until 100% done. REQUIRES an existing plan; it is NOT the entry point for a raw task. Th

• factory-explain

  Explain a just-finished run in plain English to someone who has zero context on it — what got built (before → now), which decisions were made without asking, what was deliberately skipped or is still owed, what new work turned up, and where the demo video and screenshots are. Ad-hoc and read-only; n

• factory-frontend

  Design-system-grade operating principles and procedures for any frontend/UI work — aesthetic direction, wireframing, hi-fi prototyping, tweak panels, faux product UI (hero stills, product animations, bento grids), accessibility/interaction/hierarchy audits, and a pre-ship polish pass. Adapted from T

• factory-humanizer

  Removes the cues that make prose read as AI-generated and forces a deliberate human voice, in English or Japanese. English runs a two-mode flow — Build (pin a register and a speaker before drafting) and Audit (a deterministic scanner, `scripts/unslop_text_scan.py`, with a scored/CI-gateable exit cod

• factory-implement-light

  The reduced factory pipeline for small/medium requests inside an initiative-listener loop — plan → one critical-only review gate over the plan (borderline findings get asked as a short question instead of silently noted, and the run waits for the answer) → execute → a quick recheck → factory-demo-li

• factory-implement

  The factory ENTRY POINT / driver. Pick up a single task from the tracker (by id/url, or an intake item) and drive it end-to-end to ready-for-review — autonomously, with L7-engineer rigor. Reads every comment, then runs the pipeline (plan → re-research → execute(build) → recheck → stand up infra + ve

• factory-initiative-archive

  Put an initiative on ice until you come back to it — stop its listening agent, move its channel out of the active list (a Discord archive category, or Slack's archive), and mark it archived in `factory/initiatives.md` and its identity doc, with everything that was in flight written down and the exac

• factory-initiative-listen

  Run as a persistent, per-channel conversational agent for one initiative — the loop that actually stays open all day. Takes a `<slug>` argument, reads that initiative's identity doc and watermark to resume, then watches its dedicated channel with a background Monitor that wakes the session the momen

• factory-initiative-start

  Bootstrap a new "initiative listener" — a persistent, per-channel conversational agent that will run in its own terminal, take ad-hoc feedback and requests on one Slack/Discord channel, and drive them through a light version of the factory pipeline. Interviews the user for the channel (existing or n

• factory-intake

  Intake and groom a backlog from an unstructured brain-dump or meeting transcript into prioritized, complexity-classified issues in the project's tracker (GitHub / GitLab / Jira / etc.). Reads per-project rules from a `factory/` directory at the repo root, and onboards that directory if it is missing

• factory-listener

  The one general-purpose, cross-initiative triage agent — a persistent conversational listener on its own channel that knows about every initiative in `factory/initiatives.md` and either routes a request into the right initiative's channel (for that initiative's own `factory-initiative-listen` to pic

• factory-manim-video

  Production pipeline for mathematical and technical animations using Manim Community Edition. Creates 3Blue1Brown-style explainer videos, algorithm visualizations, equation derivations, architecture diagrams, and data stories. Use when users request: animated explanations, math animations, concept vi

• factory-marketing-clips

  Find videos where credible commentators discuss a topic, transcribe the relevant windows, and cut short frame-accurate clips to pair with your own published content — an article, a paper, a post. Covers YouTube discovery (Chrome tools + yt-dlp), the auto-caption trap, Whisper two-stage timestamping,

• factory-marketing-dub

  Take a screen-recorded product demo whose dub is grammatically correct but obviously machine-translated, and produce a natural-sounding redub in a cloned voice, plus a matching short-form social post. Covers transcribing the existing dub, auditing it for translationese/register-crashes/meaning-inver

• factory-marketing-onboard

  Interactive, hands-on onboarding for the marketing pillar — sets up `factory/marketing/platforms.md`, `schedule.md`, and one file per content type under `content-types/` (plus its index). Unlike the rest of Factory's onboarding, this is not a batch of upfront questions — after a short setup intervie

• factory-marketing-post

  The generic hourly dispatcher for the marketing pillar — reads `factory/marketing/schedule.md` to decide whether (and what) to fire this tick, then executes the matching content type(s) against their platform(s), respecting per-content-type autonomy mode and a per-(platform, account) lock so two lan

• factory-marketing-recap

  Produce a recurring recap package — a researched narrative transcript, a hand-drawn Excalidraw board of clustered mini-graphics to talk over on video, an optional translated copy of that board, a per-graphic talk track, and an archived backup in the repo. Covers the PRE-production half of a periodic

• factory-marketing-video

  Edit any video by conversation. Transcribe, cut, color grade, generate overlay animations, burn subtitles, dub into another language with a cloned voice, publish to YouTube, cut shorts — for talking heads, montages, tutorials, travel, interviews, and recurring commentary/recap videos. No presets, no

• factory-onboard

  One-stop onboarding for a repo's `factory/` directory. Scans which per-project context files exist, and interviews the user to fill in ONLY the missing ones — so the same skill sets up a brand-new repo end to end AND back-fills a single new context file when a later skill/agent needs one. Use when t

• factory-plan-review

  Independent, fresh-eyes review of one finished implementation plan — the pre-execution counterpart to `factory-code-review`. Reads the plan file, the originating task/issue it's for, and the project's own `factory/code-review.md` (reused verbatim — this module has no separate tool config), then chec

• factory-plan

  Form a well-researched implementation plan for a single task and write it to a markdown file in ./tasks. Heavily researches the codebase for patterns, norms, opinionated logic, and existing utils before proposing anything. The planning step of the factory pipeline; also usable standalone. Use when t

• factory-prototype

  Before building a task for real, produce a disposable, no-corners-cut prototype of what "done" looks like — a fully designed, animated, production-shaped UI on a throwaway branch for frontend-facing work, or a concise before/after schema diagram + job DAG + per-job TLDR for backend-shaped work. Lets

• factory-qa

  The QA gate — actually use the thing you just built, as a user would, before anyone else has to. Stands the feature up, writes a list of real user scenarios, walks every one of them on the running stack, and iterates (fix → re-push → re-QA) until functionality, intent, hidden assumptions, and UI/UX

• factory-recheck

  Fresh-eyes, read-only re-review of your own just-finished implementation against the original requirements. Produces a PASS/WARN/FAIL table across correctness, bugs, duplication, conventions, docs, and security, plus a YES/NO verdict. The verification step of the factory pipeline; also usable standa

• factory-remember

  Turn a correction that just happened into a durable, concise checklist rule in the right Factory gate — plan review, code review, or QA — so the same class of defect gets caught before a human has to point at it again. Generalizes the incident into a rule, picks the earliest gate whose evidence can

• factory-reresearch

  Re-research an existing plan to find holes — missing requirements, anti-patterns, utils being re-created that already exist, functions used incorrectly. Updates the existing plan document in place rather than creating a new one. The hole-finding step of the factory pipeline; also usable standalone.

• factory-schedule-sync

  Build a high-leverage agenda for an alignment/sync meeting. Looks at recent changes, the backlog, and issues that lack context or alignment, cross-references who''s attending against ownership, and proposes a discussion list aimed at decisions and priorities — not status theater. Collaborative: you

• factory-sync-recap

  Turn a completed alignment-meeting transcript into board updates and refreshed priorities. Matches the transcript to the agenda it came from, cross-correlates what was decided to the real issues, updates them (context comments, owners, status, new/closed items), and captures the meeting's priorities

• factory-terminal-demo-video

  Build a polished, narrated demo video of a REAL interactive CLI or coding-agent session — captured in a PTY to an asciinema cast, time-warped, replayed inside a CSS "OS" window (macOS chrome, frosted glass, dock, localisable menu bar), recorded at 2560x1440, then dubbed with a voice-over and burned-

• factory-weekly-report

  Generate a weekly report as a single image — a tech-tree investment board showing where the work stands, what's in flight, what's blocked, and what shipped. Surveys the tracker and repo, curates the load-bearing work into lanes and time tiers, renders a PNG (plus a standalone HTML). Reads per-projec

• factory-weekly-rule-audit

  Review the past week's coding-agent conversations and outcomes, then write a Markdown proposal to simplify or improve agent rules. Use for weekly failure retrospectives and rule audits, not project status reports or individual code reviews.

Slash commands (1)

• /factory:factory-cli

  Run the Factory CLI (skills status, scheduled agents) from the installed plugin.

Similar plugins