README
¶
core
Shared core building blocks reused across the Lantern monorepo (server, client, CLI). Collects general-purpose data structures and algorithms that are not specific to the gRPC/transport layer.
Subpackages
| Path | Description |
|---|---|
cache/ |
Generic in-memory cache with TTL. |
collection/ |
Generic collections: priority queue (pq), set, slice helpers. |
concurrent/ |
Concurrency primitives, including a typed pub/sub (concurrent/pubsub). |
graph/ |
In-memory graph model and traversal/scoring helpers. |
graphcache/ |
Graph-specialized cache: TTL'd vertices and additive decaying edges, with an optional prefix index. The server's primary store. |
llm/ |
Backend-agnostic abstractions for single-shot, structured-output language models. |
model/function/ |
Functional-style type aliases used across the other packages. |
nlp/ |
Lightweight NLP helpers. |
These packages originated in a standalone toolkit repository and were folded into this monorepo when the projects were consolidated. They have no dependencies on the rest of Lantern and are safe to import from anywhere in this module.
Directories
¶
| Path | Synopsis |
|---|---|
|
example
command
|
|
|
collection
|
|
|
concurrent
|
|
|
pubsub
Package pubsub is an in-process publish/subscribe primitive for the Lantern core.
|
Package pubsub is an in-process publish/subscribe primitive for the Lantern core. |
|
pubsub/example
command
|
|
|
example
command
|
|
|
Package graphcache provides Lantern's in-memory graph cache: a TTL-backed vertex store plus additive, expiring directed edges.
|
Package graphcache provides Lantern's in-memory graph cache: a TTL-backed vertex store plus additive, expiring directed edges. |
|
example
command
|
|
|
Package hlc implements a Hybrid Logical Clock (HLC) suitable for stamping mutations in Lantern's leaderless full-replica replication design.
|
Package hlc implements a Hybrid Logical Clock (HLC) suitable for stamping mutations in Lantern's leaderless full-replica replication design. |
|
Package llm defines backend-agnostic abstractions for single-shot, structured-output language models.
|
Package llm defines backend-agnostic abstractions for single-shot, structured-output language models. |
|
anthropic
Package anthropic implements the llm.Model interface against Anthropic's Messages API (POST /v1/messages) using structured JSON-schema output.
|
Package anthropic implements the llm.Model interface against Anthropic's Messages API (POST /v1/messages) using structured JSON-schema output. |
|
anthropic/example
command
Command example shows the public API of the core/llm/anthropic package: build a reusable Client, bind a structured-output type with New, and call Generate to get a fully decoded value plus token usage and finish reason.
|
Command example shows the public API of the core/llm/anthropic package: build a reusable Client, bind a structured-output type with New, and call Generate to get a fully decoded value plus token usage and finish reason. |
|
gemini
Package gemini implements the llm.Model interface against Google's Gemini generateContent API using a response schema (responseMimeType=application/json + responseSchema).
|
Package gemini implements the llm.Model interface against Google's Gemini generateContent API using a response schema (responseMimeType=application/json + responseSchema). |
|
gemini/example
command
Command example shows the public API of the core/llm/gemini package: build a reusable Client, bind a structured-output type with New, and call Generate to get a fully decoded value plus token usage and finish reason.
|
Command example shows the public API of the core/llm/gemini package: build a reusable Client, bind a structured-output type with New, and call Generate to get a fully decoded value plus token usage and finish reason. |
|
openai
Package openai implements the llm.Model interface against OpenAI's Responses API (POST /v1/responses) using strict structured output.
|
Package openai implements the llm.Model interface against OpenAI's Responses API (POST /v1/responses) using strict structured output. |
|
openai/example
command
Command example shows the public API of the core/llm/openai package: build a reusable Client, bind a structured-output type with New, and call Generate to get a fully decoded value plus token usage and finish reason.
|
Command example shows the public API of the core/llm/openai package: build a reusable Client, bind a structured-output type with New, and call Generate to get a fully decoded value plus token usage and finish reason. |
|
model
|
|
|
Package mutationlog is the single source of truth for the stream of graph mutations that flows through Lantern's leaderless full-replica replication design.
|
Package mutationlog is the single source of truth for the stream of graph mutations that flows through Lantern's leaderless full-replica replication design. |
|
example
command
Command example shows the public API of the core/search package: build an inverted index, index documents, run ranked queries, and delete a document.
|
Command example shows the public API of the core/search package: build an inverted index, index documents, run ranked queries, and delete a document. |
|
relevance
Package relevance is the search-quality yardstick for core/search (#887): golden corpora with graded relevance judgments, the standard IR metrics (nDCG@10, MRR, Recall@50), and a runner that evaluates any ranking function against them.
|
Package relevance is the search-quality yardstick for core/search (#887): golden corpora with graded relevance judgments, the standard IR metrics (nDCG@10, MRR, Recall@50), and a runner that evaluates any ranking function against them. |
Click to show internal directories.
Click to hide internal directories.