Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
65 lines (54 loc) · 2.3 KB

CONTRIBUTING.md

File metadata and controls

65 lines (54 loc) · 2.3 KB
Raw

Contributing to AgentKeeper

Thanks for considering a contribution. AgentKeeper is small enough that most decisions get made quickly — open an issue first to discuss anything non-trivial.

Development setup

git clone https://github.com/Thinklanceai/agentkeeper.git
cd agentkeeper
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev,all]"
pytest -q
ruff check agentkeeper tests

Principles we hold

  1. No breaking changes to the public API. Every release since v0.1 has preserved the surface (agent.remember(content, critical=True) still works in v1.0). New features are additive.
  2. Vendor-agnostic. Anything that locks the library to a specific provider (OpenAI-only, Anthropic-only) goes in an optional extra, never in the core path.
  3. No central infrastructure. Storage stays local-first (SQLite). Cloud / managed features are downstream products, not in this repo.
  4. Tests come with the code. PRs without tests for the new behaviour get a polite ping.
  5. Narrative consistency. AgentKeeper is cognitive continuity infrastructure — not a "memory wrapper", not a "vector DB", not an "agent framework". Keep terminology aligned.

Workflow

  • One feature/fix per branch.
  • Commit messages: present-tense imperative (add semantic recaller cache, not added a cache).
  • Run pytest -q && ruff check agentkeeper tests before pushing.
  • Open a PR; CI must pass.

Project layout

agentkeeper/
  __init__.py          # public API, sync Agent
  async_agent.py       # AsyncAgent
  errors.py            # typed exceptions
  logging.py           # stdlib logger namespace
  retry.py             # backoff decorators
  cso/                 # CognitiveStateObject, Fact, identity, tiers
  cre/                 # Cognitive Reconstruction Engine
  semantic/            # Embeddings + vector index + recaller
  compression/         # Decay + consolidation + contradiction + pipeline
  translation/         # Per-provider cognitive profiles + renderers
  adapters/            # LLM provider adapters (sync + async)
  storage/             # SQLite persistence
  benchmark/           # Reproducible benchmarks
tests/                 # pytest suite
examples/              # Self-contained demos

Questions?

hello@thinklanceai.com or open an issue.