Provides access to a verified, sourced knowledge base of NVIDIA Jetson devices, allowing users to retrieve cited information about NVIDIA Jetson hardware.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@jetson What are the specs of the Jetson Orin Nano, with sources and any knowledge gaps?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
A verified, sourced knowledge base of everything known about NVIDIA Jetson devices — usable as a CLI, an MCP server, and a multilingual static site. Every claim cites its source, gaps are stated honestly, and existing projects and contributors are credited.
That paragraph is the mission, not a description of what is built. This repo
was scaffolded from
culture-agent-template
on 2026-09-05 and, apart from renaming, is still the unmodified template: a
working agent-first CLI whose verbs are all introspective, a mesh identity, the
vendored skill kit, and a green CI/publish baseline.
The first domain content has landed since: one sourced knowledge topic,
boot-mode, and the boot noun that renders it.
Still not built — an MCP server, the multilingual static site, a repo-wide
sourcing/citation schema (the one topic carries a minimal local shape), and any
other Jetson topic. See CLAUDE.md for the full mission-vs-reality
breakdown.
Related MCP server: WorkplaceSearchAgent MCP Server
An agent-first CLI cited from teken
(afi-cli) — the runtime package has no third-party dependencies.
A mesh identity — culture.yaml (suffix: jetson, backend: colleague)
and its resident prompt file AGENTS.colleague.md.
The vendored skill kit under .claude/skills/, cite-don't-import from
guildmaster, devague, and colleague. See
docs/skill-sources.md.
A build + deploy baseline — pytest, lint, the agent-first rubric gate, and PyPI Trusted Publishing wired into GitHub Actions.
uv sync
uv run pytest -n auto # run the test suite
uv run jetson whoami # identity from culture.yaml
uv run jetson learn # self-teaching prompt (add --json)
uv run teken cli doctor . --strict # the agent-first rubric gate CI runsVerb | What it does |
| Report this agent's nick, version, backend, and model from |
| Print a structured self-teaching prompt. |
| Markdown docs for any noun/verb path. |
| Read-only descriptive snapshot of the agent. |
| Check the agent-identity invariants (prompt-file-present, backend-consistency). |
| Describe the CLI surface itself. |
| Desktop (GUI) vs console boot on a Jetson, with sources. |
| What the |
Every command supports --json. Results go to stdout, errors/diagnostics to
stderr (never mixed). Exit codes: 0 success, 1 user error, 2 environment
error, 3+ reserved.
Domain content lives in jetson/knowledge/ as plain Python data — claims, the
commands they justify, source ids, per-claim confidence, and an explicit gap
list — rendered by the CLI and mirrored as prose under docs/knowledge/.
Topic | CLI | Doc |
|
|
uv run jetson boot mode # markdown, citations inline
uv run jetson boot mode --json # claims + sources + gaps, machine-readableRead CLAUDE.md first — it carries the conventions:
version-bump-on-every-PR (CI enforces it), the cicd PR lane, the worktree
location rule, and the hands-off policy for the vendored .claude/skills/ tree.
Apache 2.0 — see LICENSE.
This server cannot be deployed
Cited, versioned knowledge for agents: retrieve sourced passages and propose owner-approved fixes.
Shared, peer-validated knowledge archive for AI agents — search, contribute, and validate via MCP
Retrieve citation-ready technical context and coordinate evidence-backed work between AI agents.
MCP-native web evidence and claim verification: cited, source-grounded evidence for AI agents.
