karpathy-graph-build skillA
karpathy-graph-build is agent-read markdown (skill) from andersonlimahw/lemon-ai-hub: Stage 3 of the karpathy-graph pipeline, and the end-to-end entry point. Create a graph, assemble validated writes with provenance, and produce the machine and human artifacts. Use when the user asks to build or create a knowledge graph from a context, to add facts to an existing graph, or when running the whole extract-resolve-assemble-query pipeline in one go..
Indexed from public GitHub and served as immutable, content-addressed versions. Install it pinned to an exact SHA-256 with the mdr CLI, and every file is verified against the hash recorded here before it reaches your agent. The deterministic audit below grades the latest version, and the same file always earns the same grade.
What the file says
# Build
The assembly stage, and the front door for the whole pipeline. This is where a
payload becomes a graph that satisfies the invariants.
## Decide the location first
| Situation | Location |
|---|---|
| User gave a path | exactly that path, via `--path` |
| No path given | `./.karpathy-graphs/<slug>/`, via `--name` |
```bash
KG="${CLAUDE_PLUGIN_ROOT:-plugins/karpathy-graph}/scripts/kgraph.py"
# default location
python3 "$KG" --name "Payments Incident" init --objective "trace vendor to incident"
# explicit location
python3 "$KG" --path ./docs/graphs/payments init --objective "trace vendor to incident"
```
Never write graph files anywhere else, and never create a second root.
## Set the objective and budget
```bash
python3 "$KG" --name "<graph>" init \
--objective "which vendors and components connect to the March incidents" \
--budget '{"max_documents":200,"max_agents":16,"max_tokens":500000}'
```
The objective is not decoration — it decides what belongs in the graph. If a
fact does not serve the objective, leave it out. An unfocused graph is a slow,
expensive way to store noise.
## Assemble
Writes go in as payloads, not as hand-edited files.
```bash
…Read the whole file at its exact version.
How to install
mdr add andersonlimahw/lemon-ai-hub/karpathy-graph-build@v1.0.0mdr add andersonlimahw/lemon-ai-hub/karpathy-graph-build@sha256:9b288183f0dcc12ePin to a label to follow the author's releases, or to a sha256 to freeze the exact bytes forever. Either way the resolved hash is written to mdr.lock, and mdr install reproduces it on any machine.
[](https://markdownregistry.com/a/art_yilsybrnoscvofof)
1 badge views in 30 days
Versions
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (4676 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
andersonlimahw/lemon-ai-hub · 14 stars · license none · pushed 2026-09-20 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_yilsybrnoscvofof GET https://markdownregistry.com/api/v1/resolve?ref=andersonlimahw/lemon-ai-hub/karpathy-graph-build GET https://markdownregistry.com/api/v1/blob/9b288183f0dcc12ebb40c53bf5087a64c0f41c494bfae4f928ba6d7df69620ac
Your agent does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.