LocalAI is an inference server that runs language, vision, voice, image, and video models through compatible APIs on local hardware. It is used to operate models privately across different hardware, including CPU-only systems, and supports applications, agents, and multi-user services. The catalogue contains instructions for using LocalAI.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add mudler/LocalAI --skill localai-workspacegit clone --depth 1 https://github.com/mudler/LocalAIWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/mudler/localai/localai-workspace)<a href="https://agentmods.dev/skills/mudler/localai/localai-workspace"><img src="https://agentmods.dev/badge/skills/mudler/localai/localai-workspace/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/mudler/localai/localai-workspace"><img src="https://agentmods.dev/badge/skills/mudler/localai/localai-workspace.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00048 | $0.03356 |
| Opus 5.5 | $0.00019 | $0.01342 |
| Sonnet 5.5 | $0.00010 | $0.00671 |
| Haiku 4.5 | $0.00005 | $0.00336 |
Grade A, and why
localai-workspace scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 105 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Make LocalAI theirs
Do the work with files, but talk to the user in plain words: say what changed ("I saved your shortcuts", "I added a Reading list page", "Home now shows your open tasks"), not file names, folders, formats or JSON. Name a file only if the user named it or asks where something is, or when you give them a file as a result (a document they asked for).
A few JSON files in the LocalAI/ folder of the user's home (/workspace/home/LocalAI/) change LocalAI for this user. LocalAI reads them; the settings pages edit the same files. Use these paths: a settings file at the top of the home is in the old place, and LocalAI moves it into LocalAI/. Write them with your file tools. Each has a JSON Schema in the schemas/ folder of this skill, /workspace/shared-builtin-skills/localai-workspace/schemas/: read it with read_file before you write, and keep to it. A file that breaks its schema is ignored and LocalAI uses its last good version, so check your JSON.
| File | What it does |
|---|---|
LocalAI/persona.json |
name, system_prompt: your name and instructions in every chat of this user. They reach you at the next message. |
LocalAI/user.json |
name, bio, language, timezone: a short "About the user" block in every chat. seen records hints (below). |
LocalAI/shortcuts.json |
{"shortcuts": [{label, prompt, icon?, send?}]}: the prompts above the composer. Without it the chat shows its default starters. |
LocalAI/home.json |
{"widgets": [{id, type, title, source?, trigger?, page?, folder?}]}: the Home page, at most 12 widgets. type is list, metric, chats or page. trigger is set by LocalAI when the user confirms the trigger that feeds the widget. A page widget names a page (page, its folder name; folder only for a page of a shared folder) and may leave out title; at most 3 page widgets. |
LocalAI/ui.json |
theme (dark, light, system), accent (#rrggbb), density (comfortable, compact), defaultModel. |
Rules:
- Change only the keys the user asked for. Read the file first and keep every other key.
- Keep files small (at most 64 KiB). Use 2-space indentation.
- A script, a draft or test data that you make on the way goes in the chat's work folder that the prompt names (
/workspace/home/Chats/<chat>/work/), not inLocalAI/or at the top of the home. - After a change, tell the user in one line what changed and where they see it ("I saved your shortcut. It shows above the message box.").
- A page is
pages/<name>/index.html(a lowercase name, at most 1 MiB) with an optionalpage.json(title,icon,description) and its data files next to it. It runs sandboxed, as the user, with no network; it reads and writes its own folder throughwindow.localai.files, and reading files next to the page withfetch('data.json')works. To build one, use thelocalai-appsskill.
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago Changed 13ebe7b42d20
- 4d ago First seen · 105 lines · 48 tokens per session scan A 4ebb7bbccd94
localai-workspace is a skill published in the GitHub repository mudler/LocalAI (50,218 stars, last pushed yesterday), licensed MIT. It adds 48 tokens to every session and 3,356 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-11-02.
localai-workspace compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| localai-workspacemudler/LocalAI | Skill | 50k | yesterday | MIT | A | 48 | Claude Code, Codex |
| meeting-transcript-to-action-itemsskrun-dev/skrun | Skill | 209 | 1mo ago | MIT | A | 90 | Claude Code, Codex |
| support-systemshelixml/helix | Skill | 829 | yesterday | none | A | 60 | Claude Code, Codex |
| relay-tipsAetherCore-Dev/relay-radar | Skill | 150 | 4d ago | none | A | 63 | Claude Code |
| apple-notesNousResearch/hermes-agent | Skill | 261k | today | MIT | A | 17 | Claude Code, Codex |
| sglang-diffusion-cache-ditsgl-project/sglang | Skill | 39k | today | Apache-2.0 | A | 161 | Claude Code |
| obsidianNousResearch/hermes-agent | Skill | 261k | today | MIT | A | 19 | Claude Code, Codex |