Prompts, skills, templates, and registries
Manage reusable prompt content in Happier, attach it to agents and profiles, and install or export it to provider-native locations on your machines.
Happier groups reusable AI instructions into one product area with a single canonical library and several ways to apply or export what you save there.
From a user perspective, there are five related pieces:
- Library — your synced prompt docs and skill bundles
- Templates — reusable slash shortcuts that insert saved prompts
- System prompt additions — ordered instructions attached to coding, voice, and profile surfaces
- External assets — prompts/skills that exist on a machine or inside a project
- Registries — installable skill catalogs such as
skills.shor Git sources
The core rule is:
- Happier library is the source of truth
- external files are optional imports, installs, and exports
- Happier tracks explicit links so it can detect drift instead of silently overwriting files
The library
Open Settings → Prompts & Skills to manage your reusable content.
The library currently supports two item types:
- Prompts — markdown text blocks
- Skills — a
SKILL.mdbundle, optionally with supporting files
Use prompts when you want plain reusable instructions or templates. Use skills when you want a portable bundle that should export as a real skill directory.
Prompts
Prompts are best for:
- reusable instructions you want to insert into the composer
- reusable system prompt additions
- profile-specific or agent-specific guidance
- personal prompt snippets that should sync across devices
Each prompt can have:
- a title
- markdown content
- an optional folder
- optional tags
Skills
Skills are best for:
- reusable task instructions that should travel as a real skill bundle
- prompt content that needs supporting files such as templates or reference snippets
- assets you want to install into
.agents/skills,.claude/skills,.opencode/skills, or other provider-specific locations
Each skill can have:
- a title
SKILL.mdcontent- supporting files stored inside the same bundle
- an optional folder
- optional tags
Supporting files are useful for:
- templates referenced from
SKILL.md - additional markdown instructions
- configuration snippets
- other exported bundle files
Folders and tags
Use Folders when you want lightweight organization without duplicating prompts.
Recommended patterns:
- one folder per team or workstream
- one folder per provider or backend setup
- one folder for templates vs another for system additions
Tags are lighter-weight than folders and work well for:
codingvoicereviewresearchprofile
Templates (slash commands inside Happier)
Templates are Happier-native slash shortcuts built from saved prompts.
Open Settings → Prompts & Skills → Templates to create them.
Each template maps:
- a slash token such as
/reviewor/ship-check - to a target prompt from your library
- with a behavior:
- Insert
- Insert and send
Use templates when you want a fast reusable composer shortcut without turning the prompt into a system rule.
Template behavior
When you type a saved token in the composer:
- Happier resolves the linked prompt
- inserts the prompt content into the composer
- optionally sends immediately if the template is configured that way
Happier blocks collisions with built-in action commands, so a template token cannot silently override a core slash action.
System prompt additions
This is the user-facing name for what the implementation internally stores as ordered prompt stacks.
Open:
- Settings → Prompts & Skills → System prompt additions
You can attach prompts or skills to:
- the Coding agent
- the Voice agent
- any saved Profile
How it works
System prompt additions are:
- ordered
- individually enabled or disabled
- reusable across multiple surfaces
Happier resolves them into the session’s append/system prompt in a shared way across UI and CLI.
That means:
- starting a session from the app and from the CLI uses the same prompt-building logic
- profile-linked additions and surface-linked additions compose consistently
Recommended usage
Use coding additions for:
- code-review rules
- repo-specific style constraints
- output format expectations
Use voice additions for:
- shorter conversational style
- speech-friendly formatting
- dictation or accessibility preferences
Use profile additions for:
- environment- or endpoint-specific rules
- reusable “work profile” or “research profile” instructions
External assets
Open Settings → Prompts & Skills → External Assets when you want to inspect or manage prompt-related files that live on a real machine.
This screen is machine-aware and, for project-scoped assets, workspace-aware.
Examples include:
- portable Agent skills (
.agents/skills) - provider-native skill locations such as
.claude/skillsor.opencode/skills - provider-native command locations such as
.claude/commands
What you can do there
- discover supported external assets on a selected machine
- inspect what Happier can read from that machine or project
- import external assets into the library
What “managed by Happier” means
When you export or install a library item to a machine target, Happier stores an explicit external link.
That link records:
- which library item it came from
- which machine and scope it targets
- the external reference/path identity
- digests used for drift detection
This is what lets Happier show whether an external copy is still in sync.
Exporting library items to machines
Prompts and skills can be exported from their editor screens to external targets.
Use this when you want:
- a real provider-native file on disk
- a real skill directory in a repo
- compatibility with other tools that read those files directly
Install method: copy vs symlink
For bundle-based targets such as skills, Happier supports:
- Copy — writes an independent copy into the selected target
- Symlink — writes a Happier-managed materialized copy and links the target to it
Use symlink when you want easier updates and a single managed copy per install target. Use copy when you need a completely standalone on-disk copy.
Drift safety
Happier does not silently overwrite changed external assets.
If a linked external file or directory changes on disk after export:
- Happier detects the digest mismatch
- export, install, or delete actions fail with a conflict instead of overwriting
This is deliberate. The library and the machine copy are related, but Happier still treats local disk mutations as real changes that need review.
Registries
Open Settings → Prompts & Skills → Registries to browse installable skill catalogs.
Current registry concepts:
- built-in
skills.sh - configurable Git sources
- adapter-driven future registry types
skills.sh
Happier integrates with the actual public skills.sh registry surface instead of only scanning the CLI repo.
In practice:
- browsing the built-in
skills.shsource shows real public skills - searching uses the upstream
skills.shsearch API - opening an item shows details before you import or install it
Git sources
Git sources let you add:
- a repository URL, or
- a local checkout path
Happier scans those sources for skill bundles and exposes them through the same registry UI as built-in sources.
This is useful for:
- private team skill repos
- local experiments
- curated internal catalogs
Import vs install vs export
These terms mean different things in Happier:
Import
Bring something into the Happier library.
Examples:
- import a skill from
skills.sh - import an existing
.agents/skills/foodirectory from a machine
Install
Take a registry item, import it into the library, and also install it to a selected external machine target.
Result:
- synced library item
- external machine install
- explicit external link recorded by Happier
Export
Take an existing library item and write it to an external machine target.
Result:
- external machine copy or install created from a library item you already had
Provider support
Happier keeps the library model provider-agnostic, but external asset support depends on each provider’s file conventions.
Current examples include:
- Agent skills (
.agents) — portable skill standard - Claude — commands and skills where supported by the local file layout
- OpenCode — commands and skills where supported by the local file layout
- Gemini — skill locations where supported
- Copilot — skill locations where supported
The library itself still works even when a provider has no stable external asset format.
Common workflows
Build a reusable coding instruction
- Create a Prompt
- Put it in a folder if useful
- Add it to Coding agent system prompt additions
- Optionally attach it to a specific Profile
Create a reusable slash shortcut
- Create a Prompt
- Open Templates
- Map a token like
/reviewto that prompt - Use Insert or Insert and send
Install a skill from skills.sh
- Open Registries
- Select
skills.sh - Search or browse
- Open item details
- Choose:
- Import to library only, or
- Install to import + install on a machine target
Export a skill to a project
- Open the skill in the library
- Choose Manage external assets
- Pick machine, scope, target type, and install method
- Confirm the preview
Important things to understand
Library first
Happier is designed so you can reason about prompt content from one place first, then connect it to machines and providers as needed.
External files are real
If you or another tool edits an exported asset on disk, Happier treats that as a real external change and refuses silent overwrite.
Templates are different from system prompt additions
- Templates affect what you insert or send in the composer
- System prompt additions affect the session’s system/append prompt
Skills are bundles, not just text
If you need supporting files, use a skill bundle rather than a plain prompt.