karpathy-graph-swarm skillA
karpathy-graph-swarm is agent-read markdown (skill) from andersonlimahw/lemon-ai-hub: Fan the karpathy-graph pipeline out across parallel agents that share the graph as memory instead of copying transcripts. Use when the context is too large for one pass — hundreds of documents, a whole codebase, many independent hypotheses — or when the user asks to parallelize graph construction, run a swarm, or use multiple agents to build a graph..
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
# Swarm Scale out only when the work is genuinely parallel **and the reducer is defined before fan-out**. More agents without a reducer produces activity, not progress. The graph is what makes the swarm coherent: workers write findings as structured updates, and a synthesizer traverses the graph to combine them even though no single worker saw all the sources. ## Before fanning out Answer all six. If any answer is missing, do not fan out yet. 1. **Is the work independent?** Overlapping units mean duplicate entities and a resolution problem you created yourself. 2. **What is the reducer?** How do N worker payloads become one graph? Decide now, not after. 3. **What is the artifact contract?** Every worker returns the same shape. 4. **What are the budgets?** Concurrency, worker cap, tokens, per-worker timeout, retries. 5. **What is the dedup policy?** Same entity from two workers — merge how? 6. **What is the final gate?** What must be true for the graph to ship? ## Partition Give each worker a unit that stands alone: one document, one module, one incident, one hypothesis. Never split a single document across workers — …
Read the whole file at its exact version.
How to install
mdr add andersonlimahw/lemon-ai-hub/karpathy-graph-swarm@v1.0.0mdr add andersonlimahw/lemon-ai-hub/karpathy-graph-swarm@sha256:a25fd68956ba5410Pin 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_e7jcic2dgvxb4nsu)
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 (5050 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_e7jcic2dgvxb4nsu GET https://markdownregistry.com/api/v1/resolve?ref=andersonlimahw/lemon-ai-hub/karpathy-graph-swarm GET https://markdownregistry.com/api/v1/blob/a25fd68956ba5410cd8033c8dfdca21b562a86ae64e940c348851f8df85f626d
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.