Catalog PyPI
PACKAGE EXPLORERView in registry

0latency-cli

PyPI

Transparent PTY wrapper for Claude Code with role detection and cloud atom storage

anthropicatomscaptureclaudecliverbatimwrapper
Versions collected1
Direct dependencies3From release 0.3.0
Last observed20 Oct, 15:02 UTC

README

Version 0.3.0 documentation

0latency CLI

Verbatim CLI capture wrapper for Claude Code, Codex, Gemini CLI, and Aider. Captures all user inputs and agent outputs as role-tagged atoms, stored locally or synced to the 0Latency cloud memory platform.

Installation

pip install -e .

Usage

# Wrap Claude Code session (writes to cloud if authenticated)
0latency claude

# Wrap Claude Code with local-only storage
0latency --local claude

# Dry-run mode (show what would be captured)
0latency --explain claude

# Authenticate with 0Latency cloud
0latency login

# Check status
0latency status

# Pass arguments to Claude Code
0latency claude --print "what is 2+2"
0latency claude /path/to/project

Commands

  • 0latency claude [args...] - Wrap Claude Code session with verbatim capture
  • 0latency login - Authenticate with OAuth device-code flow
  • 0latency status - Show auth state, storage info, and sync status

Flags

  • --local - Force local-only storage (override cloud writes)
  • --explain - Dry-run mode showing what would be captured
  • --version - Show version information
  • --help - Show help message

License

MIT

Reliability Features (v0.3.0)

Production-hardened for long-running sessions and adverse conditions:

  • Crash Recovery: Zero atom loss via rolling buffer + auto-import
  • Backpressure Handling: Local queue (10K cap) with exponential backoff retry
  • Large Paste Support: UTF-8-safe chunking up to 1M+ characters
  • Long-Session Stability: Ring buffers keep RSS < 500MB in 4+ hour sessions
  • Interactive Prompts: Y/N and password prompts pass through (not captured)
  • Tool-Call Tracking: Multi-tool chains atomized with sequence metadata
  • Async Background: Non-blocking capture for long-running bash commands
  • Atom Batching: API efficiency via 10-atom batches or 2s flush

See docs/reliability.md for details and verification.

Downloads & popularity

Awaiting observations

No metric observations yet

Download counts and popularity will appear here as the collector receives them.

Versions 1

Collected releases
VersionPublishedDirect dependenciesOpen version
0.3.010 May 20263

Package advisories

0
No matching advisories collected

This describes our collected data; it is not a security assessment.

Repository

PENDING
0latency-ai/0latency-cli

Shared repository data · GitHub

License detected in repository
Not collected

Metadata history

Meaningful changes
Current collected state Current

Observed since 20 Oct, 15:02 UTC

No superseded metadata states collected. Unchanged observations do not create new revisions.

Record identity

package:6bf485e2bef08025fc823a211bc3b5fe425453d023704f4977144cb02a659284