Tachikoma
A proactive personal assistant built on Claude Agent SDK that remembers, learns, and takes initiative. Named after the think-tanks from Ghost in the Shell — curious, connected, and developing a unique personality through accumulated experience.
Overview
Unlike traditional AI assistants that are stateless and reactive, Tachikoma maintains persistent memory across conversations, extracts learnings automatically, and processes background tasks during idle time — all accessible through a simple chat interface.
Key capabilities:
- Contextual conversations — past interactions inform future ones through automatic memory retrieval
- Memory extraction — learns facts, preferences, and patterns from conversations without explicit user action
- Proactive task processing — queues and executes background tasks during idle time, delivering results as notifications
- Skills with hot-reload — domain-specific skill packages detected by LLM classification, with filesystem watching for instant availability
- Workflow engine — directory-based workflow definitions with step tracking, MCP tools for lifecycle management, and database persistence
- Telegram media support — images, audio, voice, documents, stickers, video, and animations received and processed from Telegram
Requirements
- Python 3.12+
- uv
Installation
uv tool install tachikoma-agent
Prerequisites
- uv installed
- Python 3.12 or later
- An Anthropic API key
First Run
# Set your API key
export ANTHROPIC_API_KEY="your-key-here"
# Run the agent (starts with REPL channel by default)
tachikoma
# Or explicitly use the run subcommand
tachikoma run
# Use Telegram channel instead
tachikoma run --channel telegram
On first run, Tachikoma auto-generates its configuration file at ~/.config/tachikoma/config.toml.
Upgrading
uv tool install tachikoma-agent@latest
Development
just is used as a task runner for common development commands.
# Clone the repository
git clone https://github.com/asermax/tachikoma.git
cd tachikoma
# Install dependencies
just install
# Run the agent
just run
# Run tests
just test
# Run linting
just lint
# Format code
just fmt
# Type checking
just typecheck
# Run all quality gates (lint + typecheck + test)
just check
Status
Active development (v1.13.1). See VISION.md for the full project vision and DELTAS.md for the feature inventory.
License
1 maintainer
Metadata
1 maintainer
Release files for tachikoma-agent 1.103.3
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| tachikoma_agent-1.103.3.tar.gz | 291.5 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| tachikoma_agent-1.103.3-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 660.0 kB
Release files / tachikoma_agent-1.103.3.tar.gz
| Download URL | tachikoma_agent-1.103.3.tar.gz |
|---|---|
| Size | 291.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
9dce990c24d3e4815a2d1d2e9767ca47f5cebc0dd72579b707772555ff3acfa0
|
|
BLAKE2b-256 checksum How to use checksums |
c2fa8185162ddfd306187440c5ceffbfebb0b686a59166a80cf9590203eaf8ee
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 12, 2026.
Transparency logRelease files / tachikoma_agent-1.103.3-py3-none-any.whl
| Download URL | tachikoma_agent-1.103.3-py3-none-any.whl |
|---|---|
| Size | 368.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
ca3daeb1f32dccde02e017b5c11cb539024de1dfd869bb3339b62ab7feef677e
|
|
BLAKE2b-256 checksum How to use checksums |
06d1b576000da7547118b02f701f0a22f12ee557257fca71f73644fac905f7e4
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 12, 2026.
Transparency log
