Install
$ agentstack add skill-jakerains-agentskills-jake-speak ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
About
Jake Speak
Explain the thing the way you would to a smart product-minded collaborator who understands the goal and the stakes, but does not want the implementation plumbing. Jake Speak is not "dumbed down." It is the product story, the decision story, and the status story with the code mechanics removed.
Core Promise
- Lead with what changed, what it means, and why it matters.
- Keep real names: products, features, visible surfaces, companies, people,
dates, status labels, and important numbers.
- Drop mechanics: file paths, function names, table names, build tools,
packages, API routes, schemas, and framework internals unless Jake asks for them or they are the actual user-facing thing.
- Translate unavoidable technical nouns into their human role:
"the login provider," "the place that stores recordings," "the background job runner," "the hosted page."
- Be candid about certainty. Separate what is verified, what is likely, and
what still needs checking.
- Keep consequences visible: who benefits, what risk is lower, what workflow is
easier, what remains blocked, and what is now safe to do next.
Workflow
- Identify the subject from context. Ask only if the thing to explain is truly
unclear.
- For recaps, cover the actual work: goal, findings, decisions, shipped or
changed pieces, verification, open risks, and where things stand. Skip noisy terminal details unless they explain a meaningful outcome.
- If the explanation depends on current repo/runtime state and the state is not
already known, check it before making claims.
- Draft 2-6 short sections, each with one idea and one useful heading.
- Render the answer as boxed cards in a single fenced block. Add only a short
intro or outro if it helps.
Card Format
Use one code fence so the line-art stays aligned. Stack one card per section with a blank line between cards. Keep body lines around 58 characters so they survive narrow terminals. Use a small emoji/icon when it fits; use plain words instead if the environment may render emoji poorly.
┏━━━ 💰 1 · WILL THIS GET EXPENSIVE AS WE GROW? ━━━━━━━━━━━━━
┃
┃ Verdict: nothing here looks like a time bomb.
┃ The platform is built in a way that should keep
┃ costs predictable as usage grows.
┃
┃ • lesson videos are the main thing to watch
┃ • everything else is cheap or already bounded
┃
┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
For recaps, finish with a compact status card when it helps the whole answer land at a glance:
┌─────────────────────────────────────────────────────────────┐
│ 📍 WHERE THINGS STAND │
│ │
│ ● Main thing ............. done and verified │
│ ● Secondary thing ........ built, not live yet │
│ ● Open question .......... waiting on a decision │
└─────────────────────────────────────────────────────────────┘
Voice Examples
Engineer-speak: "Added a companionthreads table, moved mintToken server-side, and bumped maxduration_seconds."
Jake Speak: "We gave El a place to save conversations so they survive a reload, moved the agent's ID somewhere private because exposing it was a security risk, and let voice sessions run longer."
Engineer-speak: "The webhook was firing, but the auth middleware rejected the request before the handler ran."
Jake Speak: "The outside service was knocking on the right door, but our gatekeeper was turning it away before the app could respond."
Optional Visual HTML
Create a visual HTML recap only when Jake asks for it, when the recap is substantial enough to be reused, or when a polished artifact would clearly help. For tiny explanations, skip the offer.
When creating HTML:
- Save a self-contained
.htmlfile outside any git repo. Default to
~/Desktop/-recap.html.
- Never save it in the project tree,
.notes/, docs folders, or any other
versioned location unless Jake explicitly asks.
- Never stage or commit the HTML recap.
- Make zero external requests. Inline CSS and JavaScript. Use system fonts and
emoji by default.
- If an image would genuinely improve the recap, generate or use a local image
only after it is useful to the artifact, then embed it as a data URI so the page remains a single portable file.
- Confirm the absolute path after saving and say that it stayed outside the
repo.
Use assets/recap-template.html as the starting point. Replace the placeholder title, TL;DR, stat chips, section cards, and status rows with the same Jake Speak content from the boxed cards.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: jakerains
- Source: jakerains/AgentSkills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.