feat: Add work item type distinction

[dvic]

Addresses #88

Changes

Adds work item type distinction to PRD skills and JSON tracker:

Prefix	Type	When to Use
US-xxx	story	User-facing features ("As a user, I want...")
TA-xxx	task	Technical work (schema, backend, bugs, maintenance)

• PRD skills now document US-xxx for stories, TA-xxx for tasks
• Adds type field to prd.json schema (inferred from prefix if not explicit)
• Beads converter uses --type=story|task flag
• Backward compatible: existing files without type field still work

Note: Haven't tested this yet - wanted to run it by you early. Happy to adjust based on feedback.

[vercel]

@dvic is attempting to deploy a commit to the plgeek Team on Vercel.

A member of the Team first needs to authorize it.

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[subsy]

@dvic I think just keeping it as story / task is the best path here. My inclination is that bug and chore types dont add much value context-wise for the model - they're both tasks. Does that work for you?

[dvic]

@dvic I think just keeping it as story / task is the best path here. My inclination is that bug and chore types dont add much value context-wise for the model - they're both tasks. Does that work for you?

Sure, I've updated the PR

[subsy]

@dvic I've been reviewing the changes and they are fine.
But I can't help thinking about the net change to the system and how it performs.
Have you done any a/b tests with the same input ie what you want to build) into the PRD creator, and assessed differences in eventual output of what gets build by the ralph loop?
Because models are so good at figuring things out, I'm wondering if the extra dimensions here is help or hindrance to the model, and perhaps just letting it figure out what it needs to do from a set of user stories is the better approach. Curious to learn about what results you've seen in testing.

[dvic]

@dvic I've been reviewing the changes and they are fine. But I can't help thinking about the net change to the system and how it performs. Have you done any a/b tests with the same input ie what you want to build) into the PRD creator, and assessed differences in eventual output of what gets build by the ralph loop? Because models are so good at figuring things out, I'm wondering if the extra dimensions here is help or hindrance to the model, and perhaps just letting it figure out what it needs to do from a set of user stories is the better approach. Curious to learn about what results you've seen in testing.

I'll try to do a mini project with and without this PR and see what the differences are :)