Continuously fresh context for AI agents

An open-source Python framework that keeps code, documents, and meeting notes ready for your agents. Reprocess only what changes.

Try a change Star on GitHub —

pip install -U cocoindex

raw · incoming

Source

code · PRs · logs · messages

repo/
api.ts Δ
#eng
notes.md
PR #482

Δ only

derived · always fresh

Searchable data

vectors · graph · tables

CocoIndex keeps this in sync

reads

reads · decides · writes

Agent

code · plans · decisions

› fix getUser null check

reads view · 4 call sites

› opens PR #483

agent output becomes the next run's source

Connect your sources and targets.

Use a built-in connector or write a custom integration in Python. Declare the data your target should hold.

All connectors →

sources

Codebases
Meeting notes
Web APIs
Files and object storage
Databases
Message queues
Images and video
Audio and transcripts

flow.py

@coco.fn(memo=True)
def index(doc, target):
    row = to_row(split(doc))
    target.declare_row(row=row)

cocoindex engine

targets

Relational DB
Data warehouse
Vector DB
Graph DB
Message queue

The engine handles the update work

Memoization
Reuse results when inputs and function code stay unchanged.
Logic tracking
The engine tracks code dependencies to identify affected work.
Source tracking
Processing components connect source items to their declared target state.
Parallel processing
Independent work runs concurrently within configured limits.

Compare Open Source and Plus capabilities and availability →

Write the transform. The engine keeps it fresh.

In this example, one file changes. Cached results skip six unchanged files, and the engine updates one Postgres row.

See the React ↔ CocoIndex mental model →

flow.py · excerptRunnable quickstart →

import cocoindex as coco
from cocoindex.connectors import postgres

@coco.fn.as_async(batching=True)
async def embed(texts: list[str]) -> list[list[float]]: ...

@coco.fn(memo=True)
async def process_file(
    file: FileLike,
    target: postgres.TableTarget[DocRow],
):
    summary = await summarize(file)
    vector = await embed(summary)
    target.declare_row(row=DocRow(
        name=file.name,
        summary=summary,
        embedding=vector,
    ))

docs_flowillustrative update

sourcedocs/

intro.md
api.md Δ
guide.md
faq.md
install.md
config.md
notes.md

Δ api.md6 unchanged, served from cache

cocoindex engineprocess_file

summarize
embed
declare_row

upsert 1 row

materialized viewpostgres · doc_rows

name	summary	embedding
intro.md	Introduction to the API	0.07 0.33 …
api.md Δ	Updated API reference	0.12 −0.08 …
guide.md	Usage guide	−0.22 0.05 …
faq.md	Common questions	0.31 −0.26 …

+ 3 rows unchanged

Source changes. Update changed items and remove target rows when their sources disappear.

Code changes. Re-run affected work when function logic changes. How change detection works →

Find code and trace what a change affects.

CocoIndex Code Plus is the licensed code search service built on CocoIndex Plus. Search code and follow symbol references across your indexed repositories.

See CocoIndex Code →

your repo

src/
flow.pyΔ
pipeline.py
embed.py
api/
route.ts
tests/
t_embed.py

ast · flow.py

reads

coding agent

› where is embed() called?

3 callers · fresh index

src/flow.py:42

src/pipeline.py:18

tests/t_embed.py:7

›

Complete functions
Syntax-aware chunks keep functions and classes together in search results.
Search by meaning
Find code by its purpose, then inspect exact text and symbols.
Trace callers
Find the callers of a changed function across indexed repositories.

Built for coding, code-review and security-review agents.

Built with CocoIndex.

Clone an example and configure its source. Each example includes a flow and setup instructions.

Real-time codebase indexing

Keep an index of your repo in sync with every commit. Give code-review and coding agents structured code context.

Meeting notes → knowledge graph

Extract people, topics, decisions and action items from notes into a live graph. Query them with your agent.

Automatic repository wiki

Use a large language model to summarize each repository and build an organization overview. Rerun to process changes.

See all examples →

Describe the flow. Claude writes the pipeline.

Describe your sources, transformations and targets. Use the CocoIndex skill to help Claude write the flow.

Try the Claude skill

claude · cocoindex

youIndex my /docs folder into Postgres.

claudewriting flow.py: chunk → embed → postgres target…

okflow.py written · running

logdocs/ → postgres.doc_chunks · in sync

What builders say.

At Unity, we use CocoIndex to incrementally index critical unstructured assets. With it, we've managed to dramatically reduce unnecessary data processing computation and remote LLM embedding calls.
CocoIndex is our “Kubernetes moment” – it empowers us to index and operate data with exceptional efficiency, keeping everything always current and context-ready for AI.

Build your first pipeline.

Start with one source and one target. The quickstart walks through it end to end.

Open the quickstart Star on GitHub

Running at scale? CocoIndex Plus →