getedgehq/top-down-comms

>-

O que é top-down-comms?

top-down-comms is a Claude Code agent skill that >-.

Funciona com~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/getedgehq/skills/tree/main/top-down-comms

Installed? Explore more Programação e Desenvolvimento skills: steipete/bluebubbles, steipete/eightctl, steipete/blucli · View all 6 →

Perguntar na sua IA favorita

Abre um novo chat com esta habilidade de agente já pré-carregada.

Documentação

Top-down client communication

Tone skills tell you how it sounds. This one tells you what goes first, and why. A perfectly-voiced document with an inverted argument still fails.

The failure this exists to stop

A reviewer, on a proposal page that had already passed three visual rounds: "This is better now, but fix the storyline. Why should we start with this? Think about this from the perspective of a McKinsey consultant, a BCG consultant, and top-down communication."

The page opened with "No automated analysis of the source imagery" — a scope exclusion. It was there for a good reason: the recipient's own experts had beaten a machine benchmark, and liability was their live objection. But a reader's first impression was a disclaimer about something the document was not, and they had to reach section three before learning what was actually being proposed.

Defensive openings feel responsible and read as weak. The instinct to front-load the caveat is exactly the instinct to override.

Rule 1: the governing thought goes first

Minto's pyramid. The first substantive sentence is the answer, not the background, not the topic, not the caveat. Everything below supports it.

OpeningVerdict
"No automated analysis of the source imagery."Exclusion. What it is not.
"Ranking incoming orders by contribution margin"Topic label. Names a subject, asserts nothing.
"At full capacity the question is not how many orders you accept but which ones. The proposal: score every incoming request automatically by expected contribution margin."The answer. Recommendation + reason, first thing on the page.

Test: if the reader stops after sentence one, do they know what you want them to do? If not, rewrite sentence one.

Rule 2: action titles, not topic labels

Every section heading states its conclusion. A reader who reads only the headings must get the whole argument. This is the single highest-leverage move and the one most often skipped.

Topic label (weak)Action title (right)
BackgroundThe bottleneck is capacity, not demand
Scoring logicFive criteria decide what an order is actually worth
ExampleTwo requests the same morning, same effort, different margin
The testCheckable in an hour, before anything is built
EffortA few days to running

The headline-only test. Strip the body. Read the titles in sequence. If they do not form a coherent argument, the structure is wrong and no amount of body copy will save it.

Rule 3: SCR, and the complication is the whole job

Situation (what both sides agree is true) → Complication (why it cannot stay that way) → Resolution (your recommendation). Most weak documents have a situation and a resolution and no complication, so the recommendation arrives unmotivated.

The complication should be arithmetic where possible: "If you work orders in arrival order rather than by margin, every thin-margin order you accept costs you a fat-margin one." That is a mechanism the reader can check, not an assertion they must accept.

Rule 4: exclusions and caveats are metadata

They belong in a scan strip near the top or a numbered section at the end. Never in the opening position, never as a heading before the recommendation.

The "at a glance" strip solves this: a four-cell row directly under the governing thought.

IN SCOPE            NOT IN SCOPE           DATA SOURCE            NEXT STEP
The order in        Automated analysis     Accounting exports,    20 to 30 invoices,
which requests      of the source          not project folders    about 1 hour
are worked          imagery

The exclusion is visible in the first screen, so nobody can claim it was buried, but it is framed as scope rather than apology. Give it a full section at the end if the objection is serious enough to deserve one.

Rule 5: number the sections

01, 02, 03. Consultants do it because it makes a document referenceable in a meeting ("go back to 02"). It also forces you to notice when you have eleven sections, which means you have no argument, only a list.

Rule 6: one falsifiable next step

End on something specific, small, and capable of failing. Not "let's discuss". The strongest closer names the kill condition:

If the ranking is wrong, the logic is wrong, and the idea is dead.

An offer that can fail is more credible than an offer that cannot, especially to a room that has already been sold to.

Rule 7: MECE, and cut to the load-bearing sections

Sections must not overlap and must not leave a hole in the argument. If two sections make the same point, merge them. If a section supports nothing above it, delete it.

Applied test from the same document: an unrelated reference case sat in the middle of a proposal about order prioritisation. The reviewer's note: "the text is random ... either it goes in the email, or if it goes here, there has to be clear context." It supported a different governing thought, so it belonged in a different artifact. It moved to the covering email and the page got shorter and stronger.

Checklist before shipping

  1. First substantive sentence is the recommendation, not background or a caveat.
  2. Headings read alone as a complete argument.
  3. There is a stated complication, ideally arithmetic.
  4. Exclusions live in a scan strip or a closing section, never the opening.
  5. Sections are numbered and MECE.
  6. The close is one specific, falsifiable next step.
  7. Nothing on the page supports a different governing thought.

Not covered here

Tone and register: this skill is about argument order, not voice. Length caps for short client messages, and the build and publish mechanics of a share page, are separate concerns.

Individual skills in this repo

This repo contains 15 individual skills — each has its own dedicated page.

getedgehq/agent-evals

Build evals for an AI agent that already does real work. Use when someone asks "how do I know my agent is right", wants to test an agent before trusting it, compare models on cost versus quality, or turn production failures into tests. Walks from first tasks and yes/no verifiers, to isolated environments, to a trace-driven improvement loop.

getedgehq/cli-ux-review

Audit a command-line tool for user-friendliness — clear situation / next-step / options in every output, colour-highlighted runnable commands, no raw jargon, no silent hangs. Invoke for "CLI UX audit", "review my CLI", "is this CLI intuitive", or before any CLI release.

getedgehq/generate-image

Generate images with OpenAI GPT Image 2 via the Codex CLI, billed through the user's ChatGPT Plus subscription (no OpenAI API key, no per-image API cost). Use when the user asks to create, generate, or make an image, picture, illustration, icon, hero graphic, or concept art from a text prompt.

getedgehq/harness-first

Diagnose and fix an unreliable, expensive, or unsafe LLM agent by auditing its harness (golden set, judge, cost caps, data layer, action approvals, tracing) before blaming or swapping the model. Use when someone says an agent is "burning tokens", "hallucinating", "brittle", gives inconsistent answers, asks whether to switch to a cheaper/better model, or wants to ship an agent or prompt change to customers.

getedgehq/http-error-triage

Run before concluding anything from an HTTP error on a third-party API. Separates a real credential/entitlement problem from a CDN or WAF block, a wrong endpoint, or a client-signature ban. Use whenever an API returns 401/403/402/429 and you are about to say "the key is dead", "credits are exhausted", "the plan lacks access", or "we are rate limited".

getedgehq/linkedin-media-prep

Convert, crop, resize, and compress images and videos for optimal LinkedIn upload quality and file size. Use when the user wants to prepare a photo, image, or video for LinkedIn (profile picture, cover photo, post image, article featured image, post video, or video ad), optimize file size for LinkedIn, crop and resize for LinkedIn dimensions, convert HEIC/RAW/PNG/MOV to LinkedIn-ready JPEG or MP4, or mentions LinkedIn media requirements, specs, or limits.

getedgehq/monid

>-

getedgehq/opendraft

An 18-agent pipeline that turns one topic line into a drafted

getedgehq/people-search

Find and rank professional people for recruiting, partnerships, sales, or research from user-provided data, public web sources, an authenticated search session, or a connected provider. Use for people discovery, LinkedIn or Sales Navigator search design, profile-list ranking, or provider filter translation.

getedgehq/product-launch-video

Turn a product URL, launch brief, or approved script into a production-ready launch video. Use for product reveals, feature announcements, SaaS launches, and narrated product films; not for generic explainers or editing existing footage.

getedgehq/rocketlist

Turn a CV into a shortlist of live startup roles from Rocketlist's public job board, including adjacent job titles the person would never have searched for, each with its published salary, the evidence for the fit, and a direct apply link. Use for "find roles I would be a strong fit for", career pivots, remote or VC-backed job hunts, and salary-visible role discovery.

getedgehq/security-audit-checklist

Run a comprehensive security audit across application code, cloud infrastructure, containers, CI/CD pipelines, and infrastructure-as-code. Covers privacy compliance, OWASP basics, secret leakage, API security, IAM misconfigurations, storage exposure, Kubernetes hardening, and network security. Use when the user asks to audit, review, or harden app or cloud security, check for secrets, scan for XSS/SQLi, verify security headers, review Terraform/CloudFormation, audit AWS/GCP/Azure configs, or perform any security-focused review.

getedgehq/shadcn-first

>-

getedgehq/strip-image-ai-metadata

Strip C2PA and AI-generation metadata from images (PNG, JPEG, WebP) to remove "Generated by AI" / "ChatGPT" / "DALL-E" labels that platforms like LinkedIn, Instagram, and X display. Use when the user wants to clean AI-generated images before posting, remove AI attribution from photos, strip C2PA manifests, sanitize image metadata for professional use, or verify whether an image contains C2PA data.

getedgehq/workplan

Create, update, or close work plans for multi-step tasks. Use when starting refactors, bug lists, feature work, migrations, or any task with 2+ steps. Also use after auto-compaction to re-orient. Triggers: "workplan", "work plan", "create a plan", "what's the plan", "where was I", or when Claude detects

Habilidades Relacionadas