Decant

Decant

Decant turns the Claude Code, Codex, Gemini CLI, OpenCode, and Cursor CLI
session logs on your machine into tangible insights. See where tokens, cost,
and agent time go, inspect context usage, find the files and tools agents
touch, and search complete transcripts from a CLI or local web UI.

Decant is local-first. It makes no outbound network calls at runtime, and your
transcripts never leave your machine. The one exception is an update check that
stays off until you turn it on. See Updating.

It does keep a copy of them. To make sessions searchable, Decant writes prompts,
tool inputs, tool output, and canonicalized raw records for retained transcript
messages into an unencrypted SQLite archive at ~/.decant/decant.db, with a
full-text index over the prompt and tool-argument text. Whatever your agents read
is in there too — source code, file contents, local paths, and any credentials
pasted into a session. Decant creates the archive owner-only (0600). If your
organization has a retention policy for agent transcripts, this archive is
subject to it. See
What the archive stores to
inspect, delete, or remove it.

Built by
Dosu,
Knowledge Infrastructure for Agents. Dosu helps make agents faster, cheaper,
and more effective.

https://github.com/user-attachments/assets/188f7cca-5b0a-4504-8eb2-a317a606d91e

Quick start

Run Decant with npx with no Bun install or global package required.

This starts the local UI at , indexes the Claude Code,
Codex, Gemini CLI, OpenCode, and Cursor CLI logs on your machine, and watches
for changes.

What you get

- One SQLite archive for Claude Code, Codex, Gemini CLI, OpenCode, and Cursor
  CLI sessions.
- Full-text search across messages, tool calls, and transcripts.
- Token, estimated cost, context, activity, tool, MCP, and
  file…