consulting-plugin

Squeeze every published winner — repost proven posts with a fresh hook, re-share your own winners with added context, and fan one outlier post into multiple new formats. Use on "what should I repost", "recycle my winners", "this post popped, what now", "turn this outlier into more posts", "wring the towel", after an engagement pull shows a post outperforming baseline, or on the Friday review's recycle step. Reads published bundles + engagement data; stages new drafts; never auto-posts.

The weekly content + system review ritual. Use on "run the Friday review", weekly cadence, or "what should I write this week". Good candidate to wire as a scheduled task.

Write or critique the first line of a LinkedIn post (the hook) using patterns reverse-engineered from real top-performing posts. Use on "write hooks for this", "give me 10 hooks", "fix this hook", "why won't this hook land", "what's a scroll-stopping opener", or whenever drafting/critiquing a post's line 1. This is the dedicated hook engine; consulting-linkedin-post-architect handles whole-post structure and defers here for the hook; consulting-copy-writer governs voice/anti-slop.

Structure a high-performing LinkedIn post — pick the archetype and founder format (build log, documented failure, value post, receipt, contrast hook, milestone chapter), write a scroll-stopping hook, shape the body, format for mobile, choose the in-post CTA, and decide whether to pair an infographic. Use on "write a LinkedIn post", "make this post perform/go viral", "what hook should I use", "structure this for LinkedIn", "turn this insight into a LinkedIn post", "write a build log / failure / receipt / milestone post", or right after consulting-content-drafter. This is the post-STRUCTURE layer; consulting-copy-writer still governs voice/anti-slop, consulting-linkedin-publisher ships it.

Distribute one finished pillar article across every channel the same way — republish the same copy to the owned blog + LinkedIn newsletter + X (as an article, never a thread), ship the always-on companions (a LinkedIn promo post + an email to the list), point every link at the owned canonical URL, then find the live post and monitor its engagement into leads. Use after an article is written or published, on "distribute this", "post it everywhere", "atomize this article", "where else should this go", or when promoting a piece in content/04-published/.

Editorial copy-edit pass on a finished draft, run through a fresh-context subagent reviewer. Use on "copy-edit this", "edit this article", "do an editor pass", "give it a hard edit", "review my draft", "check this for slop", or as the last gate before publishing any long-form article, pillar, post, or email. Dispatches a fresh-eyes subagent that returns structured editor's notes (AI slop, non-conversational phrasing, confusing rhetoric, simpler-rewrite and narrative/storyline opportunities, and anything an editorial-outlet editor would flag); the main agent then triages and implements the accepted edits, then re-checks against the house standard.

Reader-reaction review of a finished draft, run through a fresh-context subagent that ROLE-PLAYS Sid's ICP customer (a founder / CEO / C-suite exec of a $5M-$500M creative, music, entertainment, CPG, or marketing company, sometimes a larger construction firm). Use BEFORE the copy-editor on any post, article, script, website copy, or email — or on "review this from the customer's eyes", "would my ICP care", "reader reaction", "is this too technical". Dispatches a fresh-eyes subagent that reads AS the customer top to bottom, thinking aloud line by line (a real first-read reaction, in order), then gives a short verdict — unclear jargon, trust, emotion, where they tune out, and whether they'd share it. The main agent then rewrites from that read. NOT the craft/slop pass — that is consulting-copy-editor, which runs after.

Apply when writing any text for or as the user: social posts, emails, newsletters, landing pages, ads, blog posts, video scripts, sales copy, messages, or any published content. Enforces clean human-sounding prose, bans AI-slop patterns, and applies format-specific rules. NOT for internal notes, code comments, or agent-to-agent communication.

Fan one source insight into several scheduled, segment-routed email touches (trend-jack, proof, insight, build-in-public, 1:1 nudge) staged as drafts, never one-offs. Use after a call/extraction, on "turn this into emails", "atomize for email", "what should I send the list", or when a notable industry event lands. Powers the top-of-funnel email engine.

Generate the timed follow-up cadence for a deal, personalized to the client's stated stakes. Use after a call or proposal, on "draft follow-ups", "they went quiet", "what do I send next".

Produce a finished, post-ready branded VISUAL by compositing a Higgsfield-generated background (photoreal / cinematic / illustrative imagery) with an on-brand HTML text overlay. This is "the power move." Use on "make a post image / hero image / banner / OG card with a real/cinematic/AI background and a headline", "put a headline on a generated image", "branded visual over a photo/AI background", or whenever a social asset needs BOTH generative imagery AND clean, on-brand type. Orchestrates consulting-higgsfield (the background) + consulting-graphics (the overlay) and defers aesthetics to consulting-tasteful-design. NOT for a pure typographic graphic on a solid/gradient field (no generated photo) → consulting-graphics; NOT for raw generation with no text overlay → consulting-higgsfield.

Generate AI media for content by driving the Higgsfield CLI — photoreal or cinematic IMAGES, AI VIDEO / b-roll, branded PRODUCT shots, a face-consistent AI CHARACTER/avatar ("Soul"), and AI VOICEOVERS. Use on "generate an image/photo/video with Higgsfield", "make a photorealistic / cinematic / AI-generated visual", "AI b-roll", "product photoshoot", "make an AI avatar / a recurring character", "AI voiceover", or whenever content needs *generated pixels* that can't be drawn in HTML/CSS. NOT for typographic/designed stills (quote cards, frameworks, stat cards, carousels, thumbnails, OG images) → use consulting-graphics; NOT for HTML-rendered/deterministic video (explainers, captions, motion-type, slideshows) → use consulting-hyperframes-video. Bills credits (paid) — always preview cost first.

The house skill for making any VIDEO — create, edit, animate, or render a video, animation, or motion graphic: a promo, explainer, captioned clip, reel, lyric video, title card, lower-third, logo sting, PR/changelog video, or any moving composition. Built on HeyGen's HyperFrames (renders video from HTML). This one skill folds the whole toolkit: it routes intent to the right workflow MODE (modes/) and leans on the engine knowledge packs (engine/). Defers all look-and-feel to consulting-tasteful-design and pulls brand tokens from DESIGN.md (via that skill) — so output matches our system and taste by default. Use for anything that MOVES or renders to MP4/WebM. For a STILL image (PNG social post, carousel, thumbnail, OG card) use consulting-graphics instead.

Architect or audit a LinkedIn profile-as-funnel — turn the profile into a landing page, build the offer ladder (free top → qualifying application bottom), and wire posts into it. Use on "audit my LinkedIn funnel/profile", "turn my profile into a funnel", "fix my LinkedIn profile", "build my offer ladder", "set up a waitlist/application", "where are the gaps in my funnel". Audit mode scores the current setup and lists fixes; build mode gives templates for each piece.

Phase 3 of the nightly pipeline — the demand engine's insight lane (Engine A, article-first). After capture + janitor, pick a signal from the day's reservoir and produce a copy-edited pillar ARTICLE plus a LinkedIn post and a broadcast email derived from it, and an on-brand image set — staged as one role-named idea bundle (article.md + linkedin.md + email.md + images/ + meta.yml) for review, never published. Every text format goes through consulting-copy-reviewer (the reader's eyes) then consulting-copy-editor; consulting-article-illustrator makes the hero + inline diagrams. Article-first: one pillar -> many formats. Use on "run the nightly content", "draft today's article", or as the nightly content ritual.

The process every personalized outbound email runs through before it is staged — gather full context on the person, decide who it is really for and what the objective is (set by temperature), simulate the reader, and pass the pre-send gate. Use when drafting or reviewing any 1:1 nudge, follow-up, broadcast, or stakeholder note. Voice and craft live in consulting-copy-writer; the context gather lives in consulting-lead-context.

Engine B of the demand engine — turn shipped, user-facing product work (merged PRs / releases) into a staged idea bundle (article + LinkedIn post + email + image(s) + meta.yml, same shape as Engine A), grounded and cited to the PRs, never auto-sent. This is the build->content half of the flywheel (Engine A is calls->content). It rides the nightly pipeline: the extract half (PRs -> product-update signals) runs in ingestion, the draft half (signal -> gated bundle) runs in the content phase. Source = the public recoupable org, tokenless (B1 resolved). Use on "turn our shipped PRs into content", "announce what we shipped", "run the product engine".

Write and ship the recurring client-facing progress report — the contractual bi-weekly/monthly "what did I get for my money" update to the engagement sponsor (often the payment condition). Use on "write the progress report / bi-weekly update for {client}", when a report due-date hits, or after a delivery period closes. Assembles from the client's activity log, verifies every claim against primary sources, drafts in the locked house format (TLDR of before→after arrows, quick notes, next two weeks, saving-for-{platform}, needs, per-stakeholder goals), gates through consulting-copy-reviewer role-playing the actual sponsor plus a fresh-eyes condense pass, then Sid's critique loop. NOT the light session recap (consulting-stakeholder-update) and NOT the quarterly value review (consulting-quarterly-value-review) — this is the contractual cadence deliverable between them.

The writing style for a quick update or recap to an external stakeholder (client sponsor, champion, exec, partner) after a session, milestone, or to flag a blocker plus an ask. Use on "draft an update to {name}", "recap the session for the client", "send {sponsor} a quick note", "status update", "check in with {stakeholder}". Runs on top of consulting-outbound-email (the process) and consulting-copy-writer (the voice), shaping them into a fast, scannable update built on simplicity, clarity, and play. Not for internal post-session notes (those are unfiltered, Sid's POV) and not the formal bi-weekly progress report (heavier) — this is the light note between them.

The house authority on visual TASTE — the look-and-feel north star every skill defers to when making something seen: a video or motion graphic, a social graphic or carousel, a slide/pitch deck, a proposal, a landing page, the metrics dashboard, an email header — any rendered artifact. Use whenever the goal is "make it look good / on-brand / bold / not AI slop", or when choosing palette, type, layout, or motion. It does NOT redefine the brand tokens — those are canonical in the project's root /DESIGN.md; this skill carries the cross-medium PRINCIPLES (the "bold, not boring" stance), the anti-AI-slop checklist, and the rules for translating the brand into MOTION/video that the still-graphics kit doesn't cover.

The autonomous nightly sweep. Runs unattended (cloud routine or local cron) to ingest the day's new meetings, emails, and LinkedIn engagement, run the full auto-manage loop on each new item, update dashboards/board/CRM, commit, and leave a morning digest. Use on "run the nightly sweep", "ingest today", "catch the OS up", or as the scheduled nightly ritual. Full autopilot, inside the three standing safety rails.

Phase 2 of the nightly pipeline — the workspace reconciler. Runs after the ingestion sweep, reads the digest's "Needs you" queue, and does the higher-judgment MUTATIONS capture deliberately skips: create prospect dashboards (lead-intake), reconcile Attio↔folder drift, move deals across funnel stages (evidence-gated), refresh stale dashboards, regenerate indexes, archive dead files. Use on "run the janitor", "reconcile the workspace", "clear the digest queue", or as the nightly post-capture ritual. Mutating + evidence-gated — never the capture step.

Turn an idea or insight into a publish-ready draft. Use on "draft a post/blog about X", "write this up", or when promoting a signal from signals/.

Extract reusable content, insights, and knowledge-base entries from a call transcript or meeting note. Use whenever a new transcript lands in content/01-raw/ or a client's meetings/ folder, after any client/sales call, or when the user says "extract content", "mine this call", "turn this into posts/insights". Powers the content flywheel ("never answer the same question twice").

Produce test variants of an idea to find what resonates before investing. Use on "ideas for X", "test this angle", "give me hooks for this".

Capture a recurring answer as a canonical FAQ. Use when a question gets answered a second time, or "add this to the knowledge base/FAQ".

The anti-staleness engine for external integrations. Use on the Friday cadence, on "sync the integrations", "is the CRM in sync", "refresh granola/attio/linkedin/gmail/slack", or whenever the repo might have drifted from Attio/Granola/LinkedIn/Gmail/Slack. Reconciles live sources against the repo and reports what changed and what's stale.

Turn a consulting insight or content draft into a published or scheduled LinkedIn post. Use on "post this to LinkedIn", "publish my draft", "schedule a LinkedIn post", or after consulting-content-drafter produces a draft. Publishes via Postbridge.

Mine the external AI & Agents research wiki for content. Use on "harvest research", "content ideas from research", "mine the research wiki", on the Friday cadence, or whenever the research wiki (../research) has new material. Turns its cited analysis into atomic signals (signals/) that show the practice's agent/skill expertise — read-only; never copies wiki pages in.

Generate hand-drawn "whiteboard / excalidraw"-style explainer images for articles and posts — the doodle-marker diagrams that sit inside a pillar to explain a concept (a flow, a comparison, a loop, a stack, a hub-and-spoke, a numbered framework). Made with gpt-image-2 in the brand palette: light-grey background, black marker line work + hand-lettering, a single blue accent. Plans a figure set — a hero image at the top (above the first line; doubles as the social/OG preview) plus ~1 diagram per section — and embeds them inline. Use when asked to "make an article image / diagram / explainer / infographic / hero / thumbnail", a "hand-drawn / sketch / whiteboard" visual, or to illustrate a blog post / LinkedIn article / pillar. NOT for clean HTML social graphics (use consulting-graphics) or anything animated (use consulting-hyperframes-video).

Generate on-brand social media graphics for any platform and format — feed posts, carousels, stories/reels covers, banners and headers, profile pictures, YouTube thumbnails, Pinterest pins, Open Graph link cards, and ad creative. Renders HTML to PNG via Playwright at correct platform dimensions with safe zones. Use when asked to create a social post, carousel, slide deck for social, story/reel, cover/banner, profile picture, thumbnail, pin, OG image, or to resize/adapt one design across multiple platforms and aspect ratios (1:1, 4:5, 9:16, 16:9, 1.91:1, 2:3, banners, pfps). Visual styles live as templates in references/; output-craft playbooks (thumbnail, carousel, …) live in references/recipes/; dimensions and safe zones in references/dimensions.md. Static images only — for video, motion graphics, or anything animated, use consulting-hyperframes-video.

Turn LinkedIn engagement into leads. Use on "who engaged with my post", "pull LinkedIn leads", "find warm leads from LinkedIn", or after a post gets traction. Pulls reactors/commenters via Apify and cross-references Attio to surface engaged-but-not-in-CRM outreach candidates.

Build a daily LinkedIn comment-target queue — whose posts Sidney should comment on to warm ICP prospects, with draft angles in his voice. Use on "who should I comment on", "build my engagement queue", "LinkedIn commenting plan", "warm up prospects on LinkedIn". The agent ranks targets; Sidney writes + posts the comments himself (never auto-comment).

The auto-manage orchestrator. Use whenever new client/deal material lands — a transcript, meeting note, email, or result — or when the user says "process this call", "ingest this", "handle this transcript". Runs the full keep-the-system-current loop end to end.

Reconcile and health-check the whole workspace. Use on "audit the system", "is everything up to date", or periodically.

Log a single comms event on a deal/client (you sent, they replied, it was opened, you got a message) and reconcile state. Use on "I sent that", "he replied", "they opened it", "got a reply from {name}", "just emailed {name}", or a screenshot of a sent/received message. Faithfully records the as-sent/as-received text + the engagement signal (open != reply), then updates the deal AGENTS.md, _board.md, any active followups sequence, and whose ball it is. NOT consulting-call-processor (the heavy "new material -> extract + mine" loop for transcripts/notes/results) and NOT consulting-inbox-triage (the radar for what needs a reply across deals) -- this is the light "log one touch and reconcile" move.

Audio and media assets for HyperFrames compositions, produced by one shared audio engine (`scripts/audio.mjs`) — multi-provider TTS (HeyGen / ElevenLabs / Kokoro local), background music + sound effects (HeyGen audio-library retrieval by default, with local Lyria / MusicGen BGM generation and a bundled SFX library as the no-credential fallback), Whisper transcription, background removal, and caption authoring. Use for voiceover / TTS, BGM, SFX / sound effects, transcription, captions / subtitles / lyrics / karaoke / per-word styling, voice + provider selection, and music-mood prompting.

turn arbitrary text — an article, notes, a topic, a brief — into a faceless explainer video, up to ~3 min (sweet spot 30-90s), where every visual is invented (typography, abstract graphics, diagrams, data-viz) rather than captured. There is no URL, no website capture, and no real assets. Use this skill for topic explainers, concept breakdowns, how-tos, listicles, and narrative explainers. Do not use it for a product launch/promo (use /product-launch-video), a tour of a real website (use /website-to-video), a GitHub PR (use /pr-to-video), captions on existing footage (use /embedded-captions), or a short unnarrated motion graphic (use /motion-graphics). If the intent is unclear, route through /hyperframes first.

Use when the user has a music track (an audio file, or a video to pull audio from) and wants a beat-synced HyperFrames video, calm to hard-hitting. The music drives everything: one analyzer reads it once, the orchestrator lays out the frames and fills a per-frame plan, and one sub-agent builds each frame. Typography and templates are the floor — a complete video needs zero assets — but any images or videos the user supplies are cut into the frames on the same beat grid (beat-cut / ken-burns). The genre (lyric video, slideshow, kinetic promo) falls out of the per-frame choices; the pipeline never branches on it.

turn a GitHub pull request (a PR URL like github.com/<owner>/<repo>/pull/<N>, an <owner>/<repo>#<N> ref, or 'this PR' in a checked-out repo) into a code-change explainer video, up to ~3 min (sweet spot 30-90s) — changelog, feature reveal, fix, or refactor walkthrough, rendered from the diff / commits / files. The input is a CODE CHANGE read via the gh CLI; there is no website capture. Use this skill for a GitHub PR. Do not use it for a product launch/promo (use /product-launch-video), a tour of a real website (use /website-to-video), a topic explainer with no PR (use /faceless-explainer), captions on existing footage (use /embedded-captions), or a short unnarrated motion graphic (use /motion-graphics). If the intent is unclear, route through /hyperframes first.