Extend Altretta — without handing it the keys
Three doors into the same trust model. Teach your AI to work against your vault with the Altretta skill, install plugins from a signed first-party catalog, or build your own with the plugin CLI. Every one of them is signed, sandboxed and permission-scoped.
Give your AI a verifiable memory
The Altretta skill teaches an AI coding agent how to read your vault, follow the graph, and check provenance — instead of guessing. It is one command, runs against the same local MCP endpoint your desktop app exposes, and every answer it produces carries a citation and a signed anchor back to your notes.
npx skills add ApiliumCode/altretta-skill- Grounded, cited answers — never a guess
- Read-only by default; nothing changes your vault without your say-so
- Works with the AI client you already use
A first-party plugin catalog you can trust
Browse and install plugins from Altretta's official Hub — curated by Apilium, not an anonymous list. Each plugin is cryptographically signed so its authenticity can be verified before it ever runs, and it executes in a sandbox that cannot reach your filesystem, the network, or the system without a permission you grant explicitly.
- Signed at the source, verified at install
- Sandboxed runtime — no ambient access to anything
- You see and approve every permission before it runs
Build and publish your own plugin
The plugin CLI scaffolds a plugin, runs it in the same sandbox your users will see, packages it into a signed artifact, and submits it to the catalog. One tool from your first command to a published, verifiable plugin.
npx @apilium/altretta-plugin- Hot-reload locally in Developer Mode
- Package and sign into the same .akplugin format the Hub ships
- Submit to the first-party catalog
Every extension plays by the same rules
Whether it arrives as a skill, a catalog plugin or your own build, the model does not bend. Three properties hold, and they are on by default.
Signed
Each plugin is cryptographically signed, so Altretta can verify it is authentic and unmodified before installing it.
Sandboxed
Plugins run in an isolated JavaScript runtime with no unfettered filesystem, network, or system access.
Permission-scoped
A plugin declares the permissions it needs; you see and approve them before enabling, and you can review them at any time.
Ready to extend Altretta?
Install the skill, browse the catalog, or start building. All three live in the same signed, sandboxed world.