Chatlens — Development skill for Claude Code

Chatlens


**Read, search, and recover context from local AI coding sessions.**

Chatlens reads Codex, Claude Code, and Hermes conversation stores, builds a local searchable index, and turns a session into a compact work card. Use it when you remember an agent solving a problem but have lost the conversation—or when the next agent needs evidence to continue.

Python 3.11+ · Zero runtime dependencies · MIT · Local only · Early release

chatlens search 'release parser' --json
chatlens read codex:SESSION_ID --mode brief --budget 2500
chatlens card codex:SESSION_ID --json

Install

Install the versioned GitHub release in a virtual environment:

python3 -m venv .venv
. .venv/bin/activate
python -m pip install 'git+https://github.com/jonah-ux/chatlens.git@v0.1.0'
chatlens --version
chatlens --help

Alternatively, install a wheel from [GitHub Releases](https://github.com/jonah-ux/chatlens/releases). Python must provide SQLite FTS5. No API key, model account, or daemon is needed. Windows support has not been verified; CI covers Linux and macOS.

Quick start

chatlens list --json --limit 10
chatlens index --json --fail-on-error
chatlens search 'release parser' --json
chatlens read codex:SESSION_ID --mode brief --budget 2500
chatlens card codex:SESSION_ID --json

Use the source and ID from `list` or `search`. Unique bare IDs and prefixes work too; ambiguous IDs fail with candidates. Refresh the index after new conversations or exclusions. Search includes the last refresh time and coverage for each indexed source; cached results do not prove that a session is still active.

To try all three readers without using your own history:

git clone --branch v0.1.0 https://github.com/jonah-ux/chatlens.git
cd chatlens
python demos/demo.py

The demo creates temporary synthetic stores and exercises the installed package.