[MAJOR] Continuing Work on AIDE

[laz-001]

Idea

should anyone be interested in continuing work on(!) AIDE, here are some of the critical issues that need to be fixed in order to stabilize the product:

See the issues here:

https://github.com/codestoryai/aide/issues?q=is%3Aissue%20state%3Aopen%20author%3Alaz-001

To summarize:

• dev processes need to become stricter
• Minor (but highly annoying) ui flaws need to be fixed (e.g. scrolling bug)
• error-handling needs to be fine-tuned by a dedicated human developer
• contribution needs to be simplified

I estimate that stabilizing AIDE/Sidecar can be done in 1 to 3 months.

[theskcd]

ACK, on it!

[theskcd]

🛠️ Development Process Enhancement: Implemented stricter development controls and fixed UI scrolling issues

• Added: Comprehensive PR validation script with TypeScript checks, linting, tests and smoke tests
• Fixed: Auxiliary window scrolling behavior by adding proper event handling for user interactions
• Enhanced: TypeScript configuration with stricter rules for improved type safety and error catching

Ready for review - focused on improving code quality and contribution standards. @team

[matfat55]

Will they still merge prs?

[laz-001]

Will they still merge prs?

I don't think so.

[laz-001]

Will they still merge prs?

@matfat55 , notice that the team didn't even bother to add the AIDE-halted information ot the repo(s).

It's the sloppyness of some creative minds, they absolutely underperform in business/strictness related tasks.

[matfat55]

Unfortunate. Someone should make a community fork

[laz-001]

Unfortunate. Someone should make a community fork

As i wrote originally "I estimate that stabilizing AIDE/Sidecar can be done in 1 to 3 months."

this would be full-time (+ someone needs to take on marketing/sales etc. - it's a bit work). And in the end, one will pay the LLM providers, I don't know the margins.

[matfat55]

Why not suspend api providing and make it BYOK? If the co behind it dropped it and its OSS, why try to make it profitable. It doesn't breed to be commercialized, the devs who really enjoy aide will work on it themselves

[laz-001]

Why not suspend api providing and make it BYOK? If the co behind it dropped it and its OSS, why try to make it profitable. It doesn't breed to be commercialized, the devs who really enjoy aide will work on it themselves

someone (quite capable) must coordinate the efforts of all the devs. Lack of central cooridination/management was the reason that the codestory team failed.

Other than this: just start it! (either yourself, or find someone to do it)

[laz-001]

Why not suspend api providing and make it BYOK?

there's this one, don't know the quality though:

https://github.com/stackblitz-labs/bolt.diy

[matfat55]

Bolt.diy isn't an ide, it's a OS version of bolt.new which makes websites. Very impressive tool, but not like aide

[matfat55]

Ha,I wish I was capable enough! Hopefully someone better does it. If nkt, I mean there's always Zed

[laz-001]

I could do it, but it's not of any priority to me (as said, that needs full-time attention).

bolt, cursor, windsurf, zencoder, aide

settled on cursor, they've updated stuff, you don't have to confirm always the agents actions (tha's why i moved to windsurf, that's why i like AIDE much - though it went nuts sometimes doing too much).

today with cursor, I did a refactoring

• first create a plan via chat,
• then executed in an agent solo-run with just 2 confirmations from my side (after 25 steps/credits, it asks for confirmation to go on).
  • just watch the chat beautifully filling with stuff (not the chaos from AIDE), and smile, watching TV