Choosing a private model for technical documentation agents
The goal is to generate software design documents from existing templates and examples, while also checking the related code. The work includes both high-level and low- documents.
The setup would use MCP to pull data from Confluence and Jira, then connect that data to such as opencode. Opus 3.6 with Kiro-cli is already being used, but the main concern is keeping the data private instead of sending it to an outside .
The desired setup needs at least 256k context so it can handle long inputs. A key question is whether this level of reasoning really requires four 3090 s.
Key points
- The use case is creating high-level and low-level software design documents.
- The model needs to use templates, examples, code, Confluence, and Jira data.
- MCP would connect the agent to workplace data sources.
- The setup needs handling rather than relying on an external .
- The main constraints are 256k context and whether four 3090 s are necessary.