add gitingestion with a signoff message

[cnu1812]

Fixes #557, solves the #606 DCO issue

[EItanya]

This will fail

[EItanya]

I just tried this locally and this release does not exist

➜  go git:(eitanya/adk) ✗ pip install gitingest==0.4.2
Defaulting to user installation because normal site-packages is not writeable
ERROR: Could not find a version that satisfies the requirement gitingest==0.4.2 (from versions: 0.1.1, 0.1.2, 0.1.3, 0.1.4, 0.1.5)

ERROR: No matching distribution found for gitingest==0.4.2

[cnu1812]

oh, yeah, it is supported to be 0.1.2 , but will stick with the latest 0.1.5

[EItanya]

Why version 3.10?

[cnu1812]

went with Python 3.10 here cause it’s a very stable and well-supported version that works smoothly with gitingest. I tested it locally and in CI, and 3.10 avoids any potential issues that might pop up with 3.11+ . Also the workflow doesn’t need anything specific from newer versions, sticking to 3.10 felt like the safest and most predictable choice for now.

[marcinkubica]

One thing though. With this we will be saving llm digest as a release artifact - which is fine, however the general idea is gitingest is added to the branch. I'm attaching the file taken from https://gitingest.com/

kagent-dev-kagent.txt

so we'd need a run of gitingest, add digest file, commit it. So it would need to run on pr close. Depending on decision if we would like to include it.

lastly - do we go with Gitingest default filename digest.txt, or want own llmdigest.txt, gitdigest.txt ?

Also, I like the idea of saving with pipe cause it gives token count for the entire file. However it doesn't work because all gitingest is outputting is the stats. @cnu1812 do you have file somewhere? Anyways we could combine those.

❯ gitingest . > myfile.log && cat myfile.log
Analyzing source, output will be written to 'digest.txt'...
Analysis complete! Output written to: digest.txt

Summary:
Directory: <dir>
Files analyzed: 299

Estimated tokens: 379.0k

[cnu1812]

Totally agree that committing the digest file directly to the branch (rather than just attaching it to the release) is a more intentional and useful long-term approach. Running GitIngest on PR close and versioning the digest sounds great to me.

• Slight lean toward llmdigest.txt but happy to go with whatever the team prefers.

One quick question: where would we prefer to store the digest file in the repo?

[marcinkubica]

I'm using defaults, saves to digest.txt, but that's my personal preference (I can't be bothered to enter custom filename each time :) ) llmdigest.txt also sounds fine.

one thing - digest has no option to add header, and number of files and estimated tokens are useful, that's when gitingest runs

❯ gitingest 
Analyzing source, output will be written to 'digest.txt'...
Analysis complete! Output written to: digest.txt

Summary:
Directory: cmd
Files analyzed: 14

Estimated tokens: 17.8k

also, for simplicity why not have all of that as .github/actions/gitingest/action.yaml and we can call it from whatever workflow we'd choose. Also needs to "sign off" the commit (which is as much as below in the commit message)

add git ingest 

Signed-off-by: Gitingest Github Actions <gitingest@kagent.dev>