After AI Insights generates a summary, a chat input appears below the summary sections. The chat assistant has full context about your analysis run — prioritization data, complexity metrics, coupling risks, clones, and the AI Insights summary itself — so you can ask targeted, specific questions about your codebase rather than generic questions about Fortran in general.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/acdeveloper-sci/fortriage/llms.txt
Use this file to discover all available pages before exploring further.
What the chat knows
Each chat message is sent with a structured context snapshot assembled from the current analysis run. The assistant has access to:- The top 15 highest-scoring units by risk Score, including their CC (cyclomatic complexity), Fan_In, Fan_Out, LOC, implicit-none status, and equivalence usage
- The complete executive summary text produced by Forti4D for this run
- The AI Insights overview, key risks, and recommended actions already shown on screen — the chat can never contradict what you have already read
- The migration strategy distribution across all analyzed units (corrected against reachability status)
- The top common coupling risks — up to 3 COMMON block entries ranked by risk level and unit count
- The clone concentration summary — total duplicate pairs and their status distribution
The chat never receives raw Fortran source code — only aggregated analysis metrics and derived summaries. Unit names, file names, and metric values are included; the actual lines of Fortran are not. See Privacy and Data Handling for the full breakdown.
Example questions
The chat works best for targeted questions grounded in the analysis results. Here are effective prompts to try after an analysis run:- “Why is this module classified as CRITICAL?”
- “What should we refactor first?”
- “Which components are migration blockers?”
- “Explain this risk to a project manager.”
- “Compare these two risky modules.”
- “What does DEAD_CODE status mean for our project?”
- “How many units have COMMON block coupling?”
- “Walk me through the recommended actions in priority order.”
Chat context and limitations
The chat assistant is built with a focused scope, not a general-purpose AI assistant:- A system-level guardrail prompt (
chat_guardrail.md) is included with every request, keeping responses anchored to the current analysis results rather than general Fortran knowledge - The assistant does not have access to the actual Fortran source code text — only the derived metrics and summaries described above
- Conversation history is maintained for the duration of the browser session; refreshing or closing the tab clears it
- The chat is only available after AI Insights has been successfully generated for the current run — the input does not appear during the empty or error states
Rate limits and errors
The chat uses the same Gemini API key and quota as the initial insights generation step:- Each chat message consumes 1 request against your daily quota
- The free Gemini tier allows 1,500 requests per day
- Errors during chat (rate limit, timeout, network failure) display a generic error message in the chat bubble — the conversation history is preserved and you can retry
The chat feature requires AI Insights to have been generated successfully first. Run an analysis and confirm that the AI Insights summary loads before the chat input appears at the bottom of the view.