Toolbox/Agent capabilities

RAG & retrieval

retrieval and knowledge layers for agents

8 repositories, most starred first
State of the category

infiniflow/ragflow infiniflow/ragflow is the category leader; if you need a production-grade RAG engine that combines agent orchestration with template-based document understanding, start there. Otherwise pick a specialist project—some focus on deep document indexing, truly local search, semantic governance, domain workflows, or audit-safe CLIs.

Document-first indexing
Teams with large document collections need rich indexing, OCR and integration ecosystems rather than a general agent runtime, because retrieval quality and connector coverage determine accuracy at scale.
Local-first search
Organizations that require full data locality, low-latency prototypes, or tiny footprints prefer lightweight in-memory or local CLI search stacks over cloud vector services for privacy and cost control.
Semantic governance
Enterprises need reconciled metrics, approved semantic layers and authoritative metadata so agents reuse trusted values instead of fabricating SQL or KPIs.
ktx
Domain-specific workflows
Regulated or specialist domains (security, finance, etc.) benefit from curated, cross-framework mappings and workflows that encode practitioner knowledge rather than generic retrieval logic.
Auditability and safety
Production teams require reviewable, dry-run capable tooling and explicit verification for search/agent actions to meet audit, compliance and ops-safety needs.
Find your fit

Which one matches your setup?

Answer any of the questions — the shortlist updates as you go. Recommendations come from the capability passports below, nothing else.

Where should reviews happen?
Can code leave your infrastructure?
What matters most?
Model access?
Pick at least one answer to get a shortlist.
Side by side

Comparison matrix

Axes are extracted from each project's docs by our review pipeline; the maturity score is computed from stars, growth and commit activity — not an opinion. Click a column to sort.

Runs inModelsContextCost to run
90.0k +558/7d
Web appBYOKOpenAIGeminiWhole-repo analysisRe-ranking & citationsSelf-hostableYour API key (paid LLMs); self-host or use cloud
52.0k +111/7d
CLIWeb appCIBYOKOpenAIOllama / localWhole-data ingestionRerankers + filtersLocal model supportYour API key (or optional LlamaParse cloud)
35.5k +151/7d
CLIWeb appBYOKOpenAIWhole-repo analysisBasic filters/configSelf-hostableUses your LLM API key (e.g. OpenAI); cloud service may be paid.
32.1k +702/7d
CLIIDECoding-agent pluginWeb appAnthropicOpenAIGeminiIndexed frontmatter + selected skillsMatching + verificationSelf-hostableFree to clone; your model/API costs
29.2k +49/7d
CLIWeb appBYOKNo repo contextBasic filtersRuns locallyFree to self-host; optional Chroma Cloud hosted service
2.4k
CLICoding-agent pluginBYOKWhole-repo indexRanked + filtersFully localFree local use; external models may require your API key.
1.6k +7/7d
CLICoding-agent pluginBYOKAnthropicOpenAIFixed providerWhole-repo analysisDedup & human reviewSelf-hostableYour LLM API key; ktx adds no extra usage billing
1.2k +1/7d
CLICoding-agent pluginBYOKLocal-onlyDry-runs & confirmationsCloud APIs with keyVolcengine account + optional LLM API key
At a glance

Capability profiles

Six axes, 0–5 each. The shape tells you the strategy: a wide hexagon is a generalist, a spike is a specialist.

ContextNoiseCustomPrivacyModelsMaturity

Combines a production-grade RAG engine with agent orchestration and template-based document understanding to provide an agentic context layer for LLMs.

ContextNoiseCustomPrivacyModelsMaturity

A data-first RAG framework focused on document agents and agentic OCR with extensive indexing, retrieval, and 300+ integrations.

ContextNoiseCustomPrivacyModelsMaturity

Vectorless, reasoning-based hierarchical retrieval that builds a human-like table-of-contents tree for traceable, context-aware RAG without vector DBs or chunking.

ContextNoiseCustomPrivacyModelsMaturity

Maps 817 structured cybersecurity skills to six major industry frameworks (e.g., MITRE ATT&CK and MITRE F3), providing practitioner-grade, cross-framework workflows for AI agents.

ContextNoiseCustomPrivacyModelsMaturity

Provides a minimal, 4-function client API for fast in-memory prototyping plus an optional hosted Chroma Cloud for production-ready vector search.

ContextNoiseCustomPrivacyModelsMaturity

Unifies ripgrep, BM25, and vector search into a local-first search layer that serves both humans and AI agents.

Kaelio/ktx1.6k
ContextNoiseCustomPrivacyModelsMaturity

Automatically ingests wiki, semantic layers, and raw table metadata to build a reconciled context layer so agents reuse approved metrics instead of inventing SQL.

ContextNoiseCustomPrivacyModelsMaturity

Provides an automation-safe, reviewable CLI with installable "Viking skills" and explicit dry-run/confirmation/read-after-write verification for production AI search and retrieval workflows.

All repositories (8)

RAGFlow is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses cutting-edge RAG with…

90.0k
+5587d
Go
2 yrs

LlamaIndex is the leading document agent and OCR platform

52.0k
+1117d
Python
3 yrs

📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG

35.5k
+1517d
Python
1 yr

817 structured cybersecurity skills for AI agents · Mapped to 6 frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE…

32.1k
+7027d
Python
6 mo

Search infrastructure for AI

29.2k
+497d
Rust
3 yrs

Local-first search across your workspace, built for humans and AI agents.

2.4k
TypeScript
1 mo

ktx is an executable context layer for data and analytics agents 🐙 Allow Claude Code, Codex, or other AI agen…

1.6k
+77d
TypeScript
3 mo

Open CLI for integrating AI search, recommendation, and conversational retrieval into agent systems and busine…

1.2k
+17d
TypeScript
4 mo
Don't want to self-host?

Hosted alternatives

If running your own reviewer is more ops than you want, these managed services cover the same job.

Vectorize

A hosted RAG pipeline: ingest, chunk, embed and retrieve as a managed service, so you skip running the vector store and the sync jobs around it.

Try Vectorize
Pinecone

Managed vector database that underpins many retrieval stacks, run and scaled for you rather than self-hosted next to your app.

Try Pinecone
Ragie

A fully managed RAG-as-a-service API with connectors to common sources, an alternative to wiring ingestion and retrieval yourself.

Try Ragie

More in Agent capabilities