Skip to content

Develop #8

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
relydda20 merged 2 commits into from
May 13, 2026
Merged

Develop #8

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3595224
feat: LLM Gemini Service
relydda20 May 13, 2026
b4f7d7d
fix: Update .env.example
relydda20 May 13, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion .env.example
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,7 +12,7 @@ DB_PASSWORD=postgres
# AI Insights — Gemini-powered recommendations
# Insights are disabled if GOOGLE_API_KEY is unset (rest of the API still runs).
GOOGLE_API_KEY=your_google_ai_studio_key_here
GEMINI_MODEL=gemini-1.5-flash
GEMINI_MODEL=gemini-2.5-flash
INSIGHTS_ENABLED=true
INSIGHT_CHECK_INTERVAL_SECONDS=600
INSIGHT_COOLDOWN_MINUTES=45
Expand Down
4 changes: 2 additions & 2 deletions docs/ai-insights.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ A Gemini-powered pipeline that watches each user's metrics and surfaces concrete

1. The **insight scheduler** ticks every `INSIGHT_CHECK_INTERVAL_SECONDS` (default 600 s).
2. For each user with telemetry in the last `INSIGHT_ACTIVITY_WINDOW_MINUTES` (default 30), the **rule engine** evaluates today's metrics + current session.
3. If a rule fires AND the user is not in cooldown, the system asks **Gemini** (`gemini-1.5-flash` by default) to characterize the workflow state and propose one action.
3. If a rule fires AND the user is not in cooldown, the system asks **Gemini** (`gemini-2.5-flash` by default) to characterize the workflow state and propose one action.
4. Gemini returns strict JSON validated against an enum schema. Invalid responses are dropped, never persisted.
5. A `WorkflowState` row + a `Recommendation` row are written. Any prior pending recommendation for the user is marked `expired` in the same transaction.

Expand Down Expand Up @@ -63,7 +63,7 @@ All require JWT or `dvf_` API token auth. The action endpoint also verifies owne
| Variable | Default | Purpose |
| --- | --- | --- |
| `GOOGLE_API_KEY` | _(unset)_ | Required to enable the pipeline. Unset = scheduler logs a single warning at boot and disables itself; rest of the API works. |
| `GEMINI_MODEL` | `gemini-1.5-flash` | Model name passed to the SDK. |
| `GEMINI_MODEL` | `gemini-2.5-flash` | Model name passed to the SDK. |
| `INSIGHTS_ENABLED` | `true` | Kill switch. Set to `false` to disable the scheduler. |
| `INSIGHT_CHECK_INTERVAL_SECONDS` | `600` | How often the scheduler ticks. |
| `INSIGHT_COOLDOWN_MINUTES` | `45` | Minimum gap between recommendations for the same user. |
Expand Down
2 changes: 1 addition & 1 deletion src/services/llm/gemini.service.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,7 +65,7 @@ const ajv = new Ajv();
const validateInsight = ajv.compile(insightSchema);

const apiKey = process.env.GOOGLE_API_KEY;
const modelName = process.env.GEMINI_MODEL || 'gemini-1.5-flash';
const modelName = process.env.GEMINI_MODEL || 'gemini-2.5-flash';

let model = null;
if (apiKey) {
Expand Down
Loading