Discover and explore top open-source AI tools and projects—updated daily.
AlmanacCodeLiving wiki for codebases, maintained by AI agents
Top 87.8% on SourcePulse
Summary
CodeAlmanac captures essential codebase context—decisions, invariants, rationale—that static analysis misses, particularly for AI coding agents. It functions as a living wiki, storing this information as plain Markdown within the repository. This provides developers and AI agents deeper insights into system design and evolution, enhancing productivity.
How It Works
The system uses AI agents (Codex or Claude) to ingest and maintain a wiki stored in the almanac/ directory. Commands like ingest process various inputs (files, PRs, transcripts) to enrich the wiki, while garden actively improves its structure. This approach keeps the wiki version-controlled alongside code, accessible via Git and a local web viewer.
Quick Start & Requirements
Install via curl -fsSL https://codealmanac.com/install.sh | sh, uv tool install codealmanac@latest, or pip install codealmanac. Requires Python 3.12+. Initialize with codealmanac init in a repository. Interactive setup (codealmanac setup) configures AI runners (Codex default, Claude option) and automation; a non-interactive option (--yes) is available.
Highlighted Details
ingest command accepts flexible inputs: local files, Git history, GitHub PRs/issues, agent transcripts.garden command refines the wiki by fixing stale links, duplicates, and weak leads.codealmanac serve) offers read-only access with search and backlinks.almanac/ is Git-managed; auto_commit enables automated Git operations.Maintenance & Community
The README focuses on technical implementation and migration from a previous npm package. It does not detail specific maintainers, community channels, or public roadmaps.
Licensing & Compatibility
The provided README does not specify a software license. This omission may pose compatibility concerns for commercial use or integration into proprietary projects.
Limitations & Caveats
This release is strictly local-only, lacking hosted integration, cloud write paths, or public SDKs. Write operations require a configured, authenticated local AI harness (Codex or Claude), and the system mandates Python 3.12+. The absence of a specified license is a significant adoption blocker.
14 hours ago
Inactive
context-labs
yamadashy