agentleFS
allthriveaiallthriveai/amplify/.claude/skills/init/SKILL.md

init

Skill0 starsChanged 3 months agoMITView on GitHub

Interactively sets up Amplify in an Obsidian vault. Asks for the vault path, scaffolds all directories, writes .amplifyrc, then walks through the voice interview to populate Voice.md. Use when the user runs /init, wants to set up Amplify, or is starting fresh.

137 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use daily-arxiv, exp-eval and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • LambenthanSkill

    daily-arxiv

    每日从 arXiv 拉取新论文,过滤相关性,auto-ingest 高优先级论文,检测 SOTA 更新

    Also in 5 projects that use this

  • LambenthanSkill

    exp-eval

    实验判决门:Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges

    Also in 5 projects that use this

  • skyllwtSkill

    exp-pilot-eval

    Pilot result evaluation — read pilot results, apply success criteria, update idea page (pilot_result, failure_reason if failed), generate PILOT_VERDICT_REPORT. Called by /ideate Phase 5 after /exp-pilot-run.

    Also in 4 projects that use this

  • skyllwtSkill

    exp-pilot-run

    Pilot experiment execution — read Pilot Spec YAML, write pilot code, run experiment(Confirm with the user before operation and require the applicant to conduct manual inspection), return results. Called by /ideate Phase 5. Does NOT modify wiki pages or judge pass/fail.

    Also in 4 projects that use this

  • gaebalaiSkill

    codex-system

    PROACTIVELY consult Codex CLI, your highly capable supporter with exceptional reasoning and task completion abilities. Codex is a trusted expert you should ALWAYS consult BEFORE making decisions on: design choices, implementation approaches, debugging strategies, refactoring plans, or any non-trivial problem. When uncertain, consult Codex. Don't hesitate - Codex provides better analysis. Explicit triggers: "think deeper", "analyze", "second opinion", "consult codex".

    Also in 3 projects that use this

  • +6 more from Lambenthan
  • +1 more from skyllwt

The file

Show its rules and text

We split this skill into the rules it gives an agent

32 rules. 0 are common, 15 are distinctive, and 17 are very specific to this project.

  • 0

    Common

    Most of the 125 edited versions we read keep these.

  • 15

    Distinctive

    This author's own choices. Worth a look.

  • 17

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    When the user runs /init, optionally followed by a vault path:

    line 10
  2. Very specific:

    If the user provided a path (e.g.

    line 14
  3. Distinctive:

    If not, use AskUserQuestion to ask:

    line 16
  4. Very specific:

    Create these with mkdir -p:

    line 31
  5. Very specific:

    Add a README.md with # {folder-name}\n to each directory that doesn't have one —

    line 55
  6. Distinctive:

    Instead, seed the wiki's two special files:

    line 60
  7. Distinctive:

    Don't overwrite an existing CLAUDE.md — if one is there, show the user the template and ask whether to merge.

    line 93
  8. Very specific:

    If .amplifyrc doesn't already exist in the vault root, write it:

    line 98
  9. Very specific:

    If .amplifyrc already exists, read it and confirm the vault path matches.

    line 133
  10. Very specific:

    If Amplify/Memory/preferences.md doesn't exist, write:

    line 137
  11. Distinctive:

    If the user gives a real answer, use it.

    line 173
  12. Distinctive:

    If the user says "skip" or gives no answer, use the placeholder for that section.

    line 174
  13. Very specific:

    After Voice.md is written, copy the generic Amplify templates from the Amplify repo into the vault and personalize them using the user's …

    line 213
  14. Very specific:

    Use cp to copy each file. Don't overwrite files that already exist in the vault (use cp -n).

    line 231
  15. Distinctive:

    Read the Voice.md you just wrote.

    line 235
  16. Very specific:

    Do a find-and-replace across all copied Amplify files in the vault:

    line 247
  17. Very specific:

    All files in Hooks/

    line 248
  18. Distinctive:

    Only replace brackets that appear in the template files.

    line 254
  19. Distinctive:

    Ask if they want to set up their brand:

    line 262
  20. Very specific:

    If they choose to set up, run the brand interview inline (same as /brand Setup Mode Steps 1-4).

    line 270
  21. Very specific:

    If they skip, move on. They can run /brand later.

    line 272
  22. Distinctive:

    Ask if they want to set up video production:

    line 276
  23. Distinctive:

    If they choose to set up, walk through four values one at a time:

    line 284
  24. Very specific:

    If the user doesn't know their HeyGen voice ID, offer to look it up: use the HeyGen API (GET https://api.heygen.com/v2/voices with the API …

    line 295
  25. Distinctive:

    If the user provides a value, save it.

    line 304
  26. Distinctive:

    If the user says "skip", leave it empty.

    line 305
  27. Very specific:

    After collecting, update the studio section in the .amplifyrc file with the provided values.

    line 308
  28. Distinctive:

    If all four values were skipped, don't write the studio section.

    line 310
  29. Very specific:

    If any voice sections were skipped, add: "Run /voice anytime to fill in the rest." If voice sections were filled, add: "Amplify templates …

    line 329
  30. Very specific:

    If voice sections were filled, add: "Amplify templates personalized with your voice." If brand was skipped, add: "Run /brand anytime to set …

    line 330
  31. Distinctive:

    If brand was skipped, add: "Run /brand anytime to set up your visual identity." If studio was skipped, add: "Run /init again to set up …

    line 331
  32. Distinctive:

    If studio was skipped, add: "Run /init again to set up video production later."

    line 332

“Common” is measured: kept by at least 6 in 10 of the 125 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Initialize Amplify

Instructions

When the user runs /init, optionally followed by a vault path:

Step 1: Determine Vault Path

If the user provided a path (e.g. /init ~/my-vault), use it.

If not, use AskUserQuestion to ask:

"Where's your Obsidian vault? Give me the path."

Options:

  • Current directory ({cwd})
  • Home folder vault (~/obsidian-vault)
  • Other (let them type a path)

Resolve the path to an absolute path. Check if it's an existing Obsidian vault (has .obsidian/ folder) or a new location.

Step 2: Scaffold Directories

The tree encodes ownership. Four buckets, each answering "may the agent rewrite this?"

Create these with mkdir -p:

Sources/Clippings          ← raw, immutable. read it, never rewrite it.
Sources/Meetings
Sources/Audio
Sources/assets

Wiki/Summaries             ← agent-owned. rewritten and kept current.
Wiki/Concepts
Wiki/Entities
Wiki/Synthesis

Work/Stories               ← projects and pipeline output
Work/Strategy

Amplify/Hooks        ← system files
Amplify/Structures
Amplify/Signals
Amplify/Memory/sessions
Amplify/Brand
Amplify/Brand/Inspiration

Add a README.md with # {folder-name}\n to each directory that doesn't have one — except the Wiki/ subfolders. The wiki navigates itself through index.md, and a hand-maintained hub sitting next to a generated index is exactly what goes stale and starts lying about what the folder holds.

Instead, seed the wiki's two special files:

Wiki/index.md:

# Index

Every page in the wiki, one line each. Read this first when answering anything.

## Summaries

## Concepts

## Entities

## Synthesis

Wiki/log.md:

# Log

Append-only. Newest entries at the bottom. Never edit an entry already written.

Step 2b: Copy the schema

Copy templates/vault/CLAUDE.md from the Amplify repo to {vaultPath}/CLAUDE.md.

This is the most important file in the vault. It defines the three layers, the single frontmatter schema, the page shapes, the ingest and query and lint workflows, and the orphan rule. Without it an agent with vault access is a chatbot; with it, the wiki maintains itself.

Don't overwrite an existing CLAUDE.md — if one is there, show the user the template and ask whether to merge.

Step 3: Write .amplifyrc

If .amplifyrc doesn't already exist in the vault root, write it:

{
  "vaultPath": "{absolute vault path}",
  "paths": {
    "sources": "Sources",
    "wiki": "Wiki",
    "meetings": "Sources/Meetings",
    "audio": "Sources/Audio",
    "stories": "Work/Stories",
    "strategyDocs": "Work/Strategy",
    "people": "Wiki/Entities",
    "voice": "Amplify/Voice.md",
    "amplifyHooks": "Amplify/Hooks",
    "amplifyStructures": "Amplify/Structures",
    "amplifyPersuasion": "Amplify",
    "signals": "Amplify/Signals",
    "memory": "Amplify/Memory",
    "brand": "Amplify/Brand"
  },
  "studio": {
    "heygenApiKey": "",
    "heygenAvatarId": "",
    "heygenVoiceId": "",
    "elevenlabsApiKey": "",
    "elevenlabsVoiceId": ""
  }
}

Note there are no research categories. Keyword-matching sources into folders sounds tidy and doesn't survive contact with a real vault — most sources match nothing and pile up at the root. Tags in frontmatter and Wiki/index.md do that job instead.

If .amplifyrc already exists, read it and confirm the vault path matches. If it doesn't match, ask the user which to keep.

Step 4: Write preferences.md

If Amplify/Memory/preferences.md doesn't exist, write:

# Preferences

## Content Style

## Coaching

## Topics

Step 5: Voice Interview

Now walk through the voice interview. Ask these five questions one at a time using AskUserQuestion or natural conversation. Wait for each answer before asking the next.

Introduce it:

"Let's set up your voice. This shapes everything Amplify writes for you. Skip any question by saying 'skip'."

Question 1: Who I am "What's your name, what do you do, and what's your background?"

Question 2: My mission "What are you trying to accomplish? What change do you want to make?"

Question 3: My audience "Who are you talking to? What do they need?"

Question 4: What I believe "What are your core beliefs? What makes your perspective different?"

Question 5: How I talk "How do you talk? Direct? Warm? Technical? Casual? What words do you use or avoid?"

For each question:

  • If the user gives a real answer, use it.
  • If the user says "skip" or gives no answer, use the placeholder for that section.

Placeholders for skipped sections:

SectionPlaceholder
Who I am[Your name, what you do, your background. Write in first person.]
My mission[What you're trying to accomplish. The change you want to make in the world.]
My audience[Who you're talking to. What they need. What keeps them up at night.]
What I believe[Your core beliefs. The hills you'll die on. What makes your perspective different.]
How I talk[Your voice: direct? warm? technical? casual? Funny? Serious? What words do you use? What do you never say?]

Step 6: Write Voice.md

Build and write {vaultPath}/Amplify/Voice.md:

# Voice

## Who I am
{answer or placeholder}

## My mission
{answer or placeholder}

## My audience
{answer or placeholder}

## What I believe
{answer or placeholder}

## How I talk
{answer or placeholder}

Preserve the user's words exactly. Clean up grammar only if needed. Don't rewrite their personality. Run a humanizer pass only on any prose you added, not their words.

Step 7: Amplify Toolkit

After Voice.md is written, copy the generic Amplify templates from the Amplify repo into the vault and personalize them using the user's voice.

7a. Copy templates

Copy all files from the Amplify source templates into the vault's Amplify directories:

Source: {amplifyRepoRoot}/templates/amplify/
Destination: {vaultPath}/Amplify/

Copy:
- Hooks/*.md → {vaultPath}/Amplify/Hooks/
- Structures/*.md → {vaultPath}/Amplify/Structures/
- Persuasion-Glossary.md → {vaultPath}/Amplify/Persuasion-Glossary.md

The amplifyRepoRoot is the directory where Amplify source code lives (the repo containing this skill file). Resolve it from the skill's own location: {skillDir}/../../.. which gives the Amplify project root.

Use cp to copy each file. Don't overwrite files that already exist in the vault (use cp -n).

7b. Personalize with Voice.md

Read the Voice.md you just wrote. Extract these values:

BracketSource
[target audience]"My audience" section: the people they're talking to
[desired result] / [achieve specific result]"My mission" section: what they're trying to accomplish
[niche]"Who I am" section: their field or domain
[old method] / [outdated method]Infer from beliefs/mission: the thing they're replacing
[negative feeling]Common pain point for their audience (infer from context)
[positive feeling]The opposite: what success feels like
[method]Their named framework or approach, if mentioned

Do a find-and-replace across all copied Amplify files in the vault:

  • All files in Hooks/
  • All files in Structures/
  • Persuasion-Glossary.md

Replace every [bracket] placeholder with the user's real context. This is automatic, not interactive. If a Voice.md section was skipped (contains a placeholder), leave those brackets unfilled.

Only replace brackets that appear in the template files. Don't modify frontmatter, headers, or structural content.

7c. Count results

Count the total number of Amplify files copied and personalized. Store this for the Report step.

Step 8: Brand Setup (optional)

Ask if they want to set up their brand:

"Want to set up your brand? This gives your videos, carousels, and articles a consistent visual identity."

Options:

  • Set up now
  • Skip for now

If they choose to set up, run the brand interview inline (same as /brand Setup Mode Steps 1-4). This writes the .amplifyrc brand section and Brand.md.

If they skip, move on. They can run /brand later.

Step 9: Studio Setup (optional)

Ask if they want to set up video production:

"Want to set up video production? This lets Amplify produce branded videos with an AI avatar from your stories. You'll need accounts with HeyGen and ElevenLabs. Skip if you're not ready."

Options:

  • Set up now
  • Skip for now

If they choose to set up, walk through four values one at a time:

1. HeyGen API key "Paste your HeyGen API key. You can find it at https://app.heygen.com/settings under API."

2. HeyGen Avatar ID "Paste your HeyGen Avatar ID. In HeyGen, go to Avatars, click on your avatar, and copy the avatar_id from the URL or API settings."

3. HeyGen Voice ID "Paste your HeyGen Voice ID. In HeyGen, go to Voices to find your voice. If you cloned your voice via ElevenLabs and synced it to HeyGen, it will appear there with a HeyGen-specific voice ID."

If the user doesn't know their HeyGen voice ID, offer to look it up: use the HeyGen API (GET https://api.heygen.com/v2/voices with the API key) to list available voices and let them pick.

5. ElevenLabs API key "Paste your ElevenLabs API key. Find it at https://elevenlabs.io/app/settings/api-keys."

6. ElevenLabs Voice ID "Paste your ElevenLabs Voice ID. In ElevenLabs, go to Voices, click on your voice, and copy the Voice ID."

For each value:

  • If the user provides a value, save it.
  • If the user says "skip", leave it empty.
  • Validate that keys look reasonable (non-empty strings, no whitespace).

After collecting, update the studio section in the .amplifyrc file with the provided values. If a .amplifyrc already exists, merge the studio key into it without overwriting other config.

If all four values were skipped, don't write the studio section.

Step 10: Report

Give a summary:

Amplify initialized in {vaultPath}

Directories: {count} created
Config: .amplifyrc written
Voice: {filled}/5 sections filled
Amplify: {count} templates installed (8 hook types, 18 structures, persuasion glossary)
Brand: {configured|skipped}
Studio: {configured|skipped}

You're ready. Try /ingest on something you have read — the flywheel needs input before it can turn.

If any voice sections were skipped, add: "Run /voice anytime to fill in the rest." If voice sections were filled, add: "Amplify templates personalized with your voice." If brand was skipped, add: "Run /brand anytime to set up your visual identity." If studio was skipped, add: "Run /init again to set up video production later."

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.