Little Heta
heta is a local command-line knowledge base. It indexes the user's
documents — PDF, Office, images, audio, code, Markdown — answers questions
from them and from saved memory, and can store new memories.
Reach for heta whenever a task needs external knowledge from the user's
own documents, or needs to recall or save memory — instead of guessing or
grepping. Run commands with the Bash tool.
1. Check Heta is set up
Run heta status once at the start.
- Shows a model provider and KB files → Heta is ready, continue.
- Shows config missing / "not configured" → tell the user to run
heta initthemselves. It is an interactive API-key setup, so do not run it yourself.
2. Four core commands
These four cover retrieval and memory. Default to heta ask.
| Command | When to use it |
|---|---|
heta ask "<question>" |
Default. Answers from saved memory and indexed documents together. |
heta query "<question>" |
When the answer must come strictly from indexed documents. |
heta recall "<query>" |
When you want the user's personal memory (past chats, facts), not documents. |
heta remember "<text>" |
When the user states a fact, decision, or preference worth keeping for later. |
Examples:
heta ask "How does our auth flow refresh tokens?"
heta query "What does the design doc say about rate limits?"
heta recall "what did I decide about the database"
heta remember "We decided to use Postgres for the main store."
Show the user Heta's output, then add a short summary.
3. Other commands
Heta can also index files (heta insert), clean up, and toggle settings.
- For a quick list: run
heta --help. - For full usage of any command: read
COMMANDS.mdin this skill's directory (next to this file). Only read it when the user actually needs one of those commands — do not load it ahead of time.