Hi! Your project looks genuinely cool.
I’ve recently been very interested in OpenWiki. Effective HTML helped me quickly understand OpenWiki’s implementation flow and system architecture, so thank you for building and sharing it.
What I experienced
There are still two practical issues in the current experience:
- The result depends heavily on the agent executing the skills correctly, because the project itself is a skill collection.
- It can be token-intensive. Using GPT-5.6-sol, generating a single architecture diagram consumed roughly 15% of my 5× plan’s token allowance.
Collaboration idea
I’m the product lead for CogFoundry. Our product can help turn the complete Effective HTML skill collection—all six skills—into an explicit, step-by-step executable workflow, aiming for more stable and repeatable generation.
Each step or sub-skill could use an appropriate model, which may reduce costs compared with using one expensive model end to end.
We can also help package this workflow as a web-based analysis service. For example, you could offer a web experience where users submit popular open-source repositories and quickly generate architecture diagrams, making Effective HTML easier for a broader audience to use.
To help get an initial collaboration started, I can provide some model tokens. CogFoundry provides access to a broad range of recent models, so you could experiment with model selection across workflow steps.
If this direction is interesting, I’d love to discuss it in depth. You can reach me on X: @wxinshn04432263.
Looking forward to connecting!
Hi! Your project looks genuinely cool.
I’ve recently been very interested in OpenWiki. Effective HTML helped me quickly understand OpenWiki’s implementation flow and system architecture, so thank you for building and sharing it.
What I experienced
There are still two practical issues in the current experience:
Collaboration idea
I’m the product lead for CogFoundry. Our product can help turn the complete Effective HTML skill collection—all six skills—into an explicit, step-by-step executable workflow, aiming for more stable and repeatable generation.
Each step or sub-skill could use an appropriate model, which may reduce costs compared with using one expensive model end to end.
We can also help package this workflow as a web-based analysis service. For example, you could offer a web experience where users submit popular open-source repositories and quickly generate architecture diagrams, making Effective HTML easier for a broader audience to use.
To help get an initial collaboration started, I can provide some model tokens. CogFoundry provides access to a broad range of recent models, so you could experiment with model selection across workflow steps.
If this direction is interesting, I’d love to discuss it in depth. You can reach me on X: @wxinshn04432263.
Looking forward to connecting!