RELEASE Posts
This release to @davaux/pka fixes some bugs regarding the CLAUDE.md (and other agent files), such as binary files not being recorded correctly or at all and issues recording the available NPM scripts. It adds new checks to identify tech stacks, including the addition of Auth and State Management, along with more libraries, and it also now records more useful info from the package.json, such as whether it's running as a module.
I don't plan to release a new version every day :P I had planned to work on more features and try to cut a new release next month, but I decided to go ahead and fix the bugs; the improvements had already been added before I found the bugs. Hopefully, I've found most of the bugs so now I can just actively use it and keep adding new features organically as the need arises, instead of such frequent releases.
I've released version 1.0.1 of my @davaux/pka NPM package. This update adds config file generation with an init command. I've also introduced config versioning, so hopefully as the config options change over time, the init function can be used to update a config file, without overriding existing settings. The init command also accepts flags to set settings to desired values when generated.
I've released version 1.0.0 of my Project Knowledge Analyzer (@davaux/pka) on NPM. This is a complete rewrite of the previous 0.9.x versions. PKA functionality has been expanded to support multiple modes and many configurable options. It can flatten a file-base for an AI like Claude Projects. It also has modes and options for AI agents, such as Claude Code, Copilot, Cursor, others, and a generic AGENTS.md file. There is also a full mode that generates everything, with configurable options to disable individual pieces. This version also adds support for a pka.config.json file in projects and is geared more toward being installed as a dev dependency, although it can still be installed globally.
There are many more new features, like chunking and setting limitations for token limits, including the stack, git history, diff summary, a watcher for auto-regeneration (useful for when working with a local agent) and more. As far as improvements to the original functionality, it now has support for file clean up upon regeneration, it excludes the output folder automatically, fixes for use on Windows, resolving imports better, and it defaults to the current directory, so you don't have to include a . on every command run.