Under the Hood
How MUJ FISH works, what it can touch, and how it stays out of SLCM's way.
1. Direct communication
MUJ FISH has no server. When it needs the details of a class, your browser asks SLCM directly, using the login session you already have. There are no proxies, no relays, and no copies of your data anywhere else.
Most of the extension makes zero network requests. It only reads the HTML table already rendered on your screen.
2. It never gets in the way
- It waits for idle time. Changes to the page are scheduled during browser idle moments
(
requestIdleCallback) or inside debounced observers, so portal rendering is never blocked. - Runaway protection. If the portal enters rapid re-render cycles or table sorting loops, a built-in guard temporarily pauses extension passes to prevent infinite DOM loops.
- No background lag. Mathematical calculations and fetches run asynchronously in small coalesced batches, keeping scrolling and clicking smooth.
3. Kind to the university's servers
These are the requests the extension makes, page by page:
| SLCM page | What it adds | Requests |
|---|---|---|
| Attendance Summary | Target stepper, safe skips, and classes needed (with exact ceiling math) | 0 |
| Internal Marks | Assessment totals, course credits, and GPA calculator mode | 0 |
| CGPA & GPA | Interactive multi-semester CGPA forecasting chart | 0 |
| Grades | Saves letter grades locally for calculator use | 0 |
| Timetable | Color-coded class status (Absent, Not Considered, Today highlight) | 0 for the list, plus 1 per completed class, then cached |
| Dashboard | Daily schedule, goal-aware “If missed” indicators, and attendance summary | 1 for the schedule, plus 1 per completed class (usually 3 to 5), then cached |
The safety limits
- Network-aware concurrency. The extension senses connection quality and dynamically limits simultaneous requests (up to 3 on high-speed connections, scaling down on slower networks), with a strict 15-second timeout.
- Deduplication. If two components request the same class simultaneously, the duplicate in-flight request is merged so the server is queried only once.
- Session caching with LRU eviction. Previously fetched classes are cached locally for the session. Revisiting dates you have already seen costs 0 requests, and cache sizes are capped to prevent memory bloat.
- Cancels outdated requests. When rapidly navigating dates or switching views, previous incomplete requests are aborted immediately.
- Backs off on errors. If 3 consecutive requests fail, automatic background fetching stops for the session to prevent server spam.
- Only checks what changes. Confirmed past classes are never re-fetched. Only recent or volatile classes are re-checked periodically.
4. Graceful failure
Every script is wrapped in defensive error handling. If SLCM alters its layout, returns an unexpected session error, or times out, the extension idles silently without crashing the page or showing disruptive popups.
If the extension updates while a page is open, the previous instance disconnects cleanly and unbinds its observers to avoid stale listeners.
Permissions, in plain words
| Permission | Why |
|---|---|
mujslcm.jaipur.manipal.edu |
To display calculations and visual indicators on that single site. It is completely inactive on any other website. |
storage |
To remember your chosen target percentage, view toggles, and predictor edits locally on your device. |
It requests no background scripting, tab monitoring, or web request interception permissions, and has zero presence on the portal login page, so your password and credentials cannot be seen or touched.
Don't take my word for it
You can check every claim in about a minute:
- Open SLCM, press
F12and go to the Network tab. - Use the portal normally. Every request stays strictly between your browser and
mujslcm.jaipur.manipal.edu. - On Attendance, Marks, or CGPA pages, you will see zero network requests fired when calculations appear.
The full source is on GitHub. For details on local data storage, see the Privacy Policy.