grepticon/
Sign inGet started

Agents don't need a vector database.
They need a filesystem.

Give your agents retrieval over isolated, permissioned workspaces. No sandbox to boot, no embeddings to maintain.

Start buildingRead the docs
Works withTypeScript SDKVercel AI SDKREST /v1MCP adapterSoongrepticon CLISoon

Four read tools over a real filesystem, hosted.

Fast enough to call in a loop, scoped to one tenant, logged end to end.

No sandbox to provision before the first read

Reads hit a hot, per-workspace index. Nothing to boot, nothing to mount, no first-call penalty.

Read the quickstart

Per-session tokens, scoped to one tenant

What a token cannot reach is absent from every listing, glob, and grep hit, not denied.

How access control works

Every read on an append-only audit log

Which token, which verb, which argument, and what came back. Filter it or export it.

See the audit API

The four unix verbs agents already know

ls, find, cat, grep. No query language, no schema, no framework glue.

Browse the tool reference

A filesystem beats a vector database for agent retrieval.

+16.7 points over naive RAG across an 18,340-file corpus, graded blind.

Retrieval accuracy
grep-agent97.9%
agentic-RAG90.6%
naive-RAG81.3%
+16.7 pts vs naive RAG · Vela corpus
Benchmark · July 23, 2026

Agent filesystem beats naive RAG on retrieval accuracy

97.9% of held-out questions answered correctly, 16.7 points ahead of naive RAG. Here is how we ran it.

5 min readRead →
Three backends, one interfaceChromaSQLiteFiles/ls · find · cat · grep
Perspective · July 24, 2026

Everyone is building a filesystem for their agent

Mintlify, Anthropic, and Turso each built one independently. Here is why the same primitive keeps reappearing.

6 min readRead →
All posts →
Use case

Integrate AI to your docs page in minutes

Turn your Fumadocs into context for an AI agent integrated directly on your docs page powered by Grepticon.

Two files · About ten minutes · No index to rebuild

Read BlogSee Guide
One question, three reads
grep"rate limit"
finddocs/**/*.md
catdocs/limits.md
No embeddings · no index to rebuild
For developers

A TypeScript SDK that drops into your agent loop.

Four read tools over the /v1 API, ready to hand to the Vercel AI SDK.

acme-docs/
├─ policies/
│  ├─ returns.md        grep "refund" ×3
│  └─ privacy.md
├─ support/
│  └─ faq.md            cat
└─ contracts/
   └─ msa.md            find **/*.md
agent.tsTypeScript
import { anthropic } from '@ai-sdk/anthropic';
import { GrepticonClient } from '@grepticon/sdk';
import { createVfsTools } from '@grepticon/sdk/ai-sdk';
import { generateText } from 'ai';
 
const client = new GrepticonClient({
  apiKey: process.env.GREPTICON_API_KEY,
});
 
const { text } = await generateText({
  model: anthropic('claude-sonnet-5'),
  tools: createVfsTools(client.session('acme-docs')),
  prompt: 'What does our refund policy cover?',
});
Read the docsBrowse the SDK
Start now

Give your agents a filesystem.

Create a workspace, push your files, and your agents can ls, find, cat, and grep in minutes. Free to start.

Start buildingRead the docs