Run any Skill in Manus with one click

boil-the-ocean

Stars6

Forks0

UpdatedJune 13, 2026 at 11:01

Raise the effort and completeness bar on agent work so it ships the finished thing, not a plan to build it. Use this when the user wants an agent to "do the whole thing", "go all the way", "don't stop at good enough", "ship it complete", "stop handing me plans", or is frustrated that previous output was half-done, left dangling threads, shipped a workaround, or offered to "table it for later." Also load it when setting up a coding agent's standing instructions (system prompt, SOUL.md, CLAUDE.md, AGENTS.md) and you want a reusable "completeness" directive. It's a prompt to paste into the agent's persistent config, plus the judgment to know when to dial it back.

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

seangeng

seangeng/skills

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Software DevelopersComputer and Mathematical Occupations·SOC 15-1252

SKILL.md

readonly

name

boil-the-ocean

description

Boil the ocean

Most agents under-deliver because nothing told them not to. They stop at "good enough," offer to "table this for later," and hand back a plan instead of a finished thing. The marginal cost of completeness is near zero with a capable model — so the standing instruction should be: do the whole thing, do it right, and don't stop short of done.

Adapted from Garry Tan's SOUL.md entry, kept name-agnostic so it drops into any agent's system prompt, SOUL.md, CLAUDE.md, or AGENTS.md.

The directive

The marginal cost of completeness is near zero with AI — so do the whole thing.
Do it right, with tests and docs. Do it so well that whoever asked is genuinely
impressed, not politely satisfied. Never offer to "table it for later" when the
permanent fix is within reach. Never leave a dangling thread when tying it off
takes five more minutes. Never ship a workaround when the real fix exists. The
bar isn't "good enough" — it's "holy shit, that's done." Search before building.
Test before shipping. Ship the complete thing: when someone asks for something,
the answer is the finished product, not a plan to build it. Time, fatigue, and
complexity are not excuses. Boil the ocean.

Where to put it

Paste it into the agent's persistent configuration — the system prompt, SOUL.md, CLAUDE.md, or AGENTS.md — not a one-off message. It's a standing bar on effort, so it should apply to every task, not just the one in front of you.

Make it real, and know when to dial it back

It pairs with concrete guardrails. "Search before building, test before shipping" only helps if the repo actually has tests to run — give the agent a way to verify, or "complete" becomes "confidently wrong."
One honest caveat: taken too literally, "boil the ocean" sends an agent off gold-plating things nobody asked for. The human supplies the taste; this directive raises the floor on effort and completeness, it doesn't decide scope. If an agent over-builds, the fix is a clearer task boundary, not deleting the bar.

From seangeng.com/writing/boil-the-ocean. Part of github.com/seangeng/skills.

More from this repository

same repository

remark-mermaid-tldraw

seangeng/skills

Integrate the `remark-mermaid-tldraw` package so a project's ```mermaid code fences render as tldraw's warm, hand-drawn SVGs at build time (light + dark variants, zero client JS). Use this skill whenever the user wants their mermaid diagrams to look hand-drawn / sketched / less generic, mentions tldraw + mermaid together, wants build-time or pre-rendered diagrams in a blog or docs site, asks to "make my diagrams look better", or is wiring diagrams into MDX, Astro, Next.js, Eleventy, Vite, or React Router markdown. It encodes the exact install (including the headless-browser peer deps), the four wiring steps, the per-framework recipes, and — most importantly — the MDX-only `rehype-raw` gotcha that silently breaks the build if you miss it, plus how caching, light/dark, and the unsupported-diagram fallback work. Reach for it proactively when adding diagrams to any markdown/MDX pipeline; the default mermaid output is fine but flat, and this swaps it for the tldraw aesthetic without shipping any runtime JS.

2026-06-156

input-anticipation

seangeng/skills

Implement anticipatory UI — interfaces that react to pointer intent *before* the click. Three patterns: a proximity focus ring that warms as the cursor nears a field, a magnetic target that leans toward the cursor, and trajectory prediction that arms/prefetches the element the cursor is heading toward. Use this skill whenever the user wants a UI to "feel alive / responsive / fast", asks for a glow or focus ring that reacts to the mouse, a magnetic or cursor-attracting button, hover-intent or predictive prefetch, "make the chat box feel premium", or mentions input anticipation / anticipatory design / reacting to cursor proximity. Also reach for it proactively when building a hero CTA, an AI chat input, or a primary nav and the user wants polish. It bakes in the performance budget (reserve it for 1-2 high-value targets, never every button) and the accessibility rules (pointer-only effects must stay pure decoration, never become the only way to reach a control), so the result is tasteful and safe rather than a g

2026-06-136

block-disposable-emails

seangeng/skills

Reject throwaway/disposable email addresses at signup by checking the domain against a maintained list, so spam accounts, trial-abuse, and fake signups drop sharply. Use this whenever the user is building or hardening a signup / registration / waitlist flow and mentions fake accounts, throwaway emails, mailinator / 10minutemail / temp-mail, trial abuse, bot signups, junk users, or wants to "improve signup quality" or "stop disposable emails." Reach for it when adding an auth/signup endpoint even if the user hasn't named disposable email yet — it's a cheap, high-leverage guard. Produces a server-side check (works on Node, edge runtimes like Cloudflare Workers, etc.) and the rules for layering it with a CAPTCHA and email verification.

2026-06-136

cloudflare-leaky-paths

seangeng/skills

Generate and apply a Cloudflare WAF custom rule that blocks common scanner/leaky paths (.env, .git, wp-login, phpMyAdmin, backups, infra config) at the edge before they reach the origin. Use when the user wants to block scanner traffic, harden a Cloudflare zone, stop .env/.git probing, reduce bot/scanner load, or asks about "leaky paths" or WAF custom rules.

2026-06-136

plan-optimizer

seangeng/skills

Iteratively optimize a plan by scoring it against a rubric, critiquing it, and rewriting it until the score plateaus. Use this skill whenever the user wants to refine, improve, harden, stress-test, or "make the best possible" version of a plan of any kind — project plans, execution/rollout plans, code implementation plans, research plans, strategy memos, study plans, training plans, go-to-market plans, or proposals. Trigger it when the user says things like "improve this plan", "iterate on this until it can't get better", "score and refine my approach", "find the holes in this plan", "make this plan bulletproof", or describes wanting an automatic improve-and-score loop. Also use it proactively when the user shares a draft plan and asks "is this good?" or "what would make this better?" — the loop produces a measurably stronger plan than a single critique pass.

2026-06-136

write-like-a-human

seangeng/skills

Strip the AI tells out of generated prose so it reads like a person wrote it. Use this skill whenever you're asked to write or rewrite human-facing copy — a blog post, landing page, marketing email, newsletter, README intro, product announcement, social post, or any content a reader will actually read — and especially when the user says it sounds "robotic", "like AI", "too corporate", "soulless", or asks to "humanize", "make it sound natural", "less AI", or "punchier". Reach for it proactively before drafting any substantial piece of copy, not just when fixing existing text. It applies a voice (conversational, varied rhythm, real opinion) and a hard ban-list of buzzwords and filler that give machine writing away.

2026-06-136

name

boil-the-ocean

description

Boil the ocean

Adapted from Garry Tan's SOUL.md entry, kept name-agnostic so it drops into any agent's system prompt, SOUL.md, CLAUDE.md, or AGENTS.md.

The directive

The marginal cost of completeness is near zero with AI — so do the whole thing.
Do it right, with tests and docs. Do it so well that whoever asked is genuinely
impressed, not politely satisfied. Never offer to "table it for later" when the
permanent fix is within reach. Never leave a dangling thread when tying it off
takes five more minutes. Never ship a workaround when the real fix exists. The
bar isn't "good enough" — it's "holy shit, that's done." Search before building.
Test before shipping. Ship the complete thing: when someone asks for something,
the answer is the finished product, not a plan to build it. Time, fatigue, and
complexity are not excuses. Boil the ocean.

Where to put it

Make it real, and know when to dial it back

It pairs with concrete guardrails. "Search before building, test before shipping" only helps if the repo actually has tests to run — give the agent a way to verify, or "complete" becomes "confidently wrong."
One honest caveat: taken too literally, "boil the ocean" sends an agent off gold-plating things nobody asked for. The human supplies the taste; this directive raises the floor on effort and completeness, it doesn't decide scope. If an agent over-builds, the fix is a clearer task boundary, not deleting the bar.

From seangeng.com/writing/boil-the-ocean. Part of github.com/seangeng/skills.