Grader

Compile all sources, run test cases in a second step, then score the rubric. Three result tables, three CSV exports.

Grading runs only on the companion server you connect to below (python3 server.py on your machine or a lab host). The browser triggers jobs and shows results; it does not compile or execute C. Put submissions where that server’s config.json points (usually next to server.py). Recommended: after starting the server, open the Grader at http://127.0.0.1:5001/ (same machine as the API — avoids blocked requests).

Paths

Submissions folder

Build output folder (gcc -o here; exe = same name as .c without extension)

Rubric JSON

Compile report CSV

Execution report CSV

Rubric report CSV

ID extraction

Regex pattern

LLM Grading (for rubric items of type "llm")

Provider

Model

API Key (saved to config.json on the server only)

Test cases & execution marks (Run execution — uses binaries from compile)

Use file-based test suites (three sets: set_1, set_2, set_3) Each student is assigned a set from their ID (stable). Edit stdin / expected per set on the same machine as server.py.

Suite assignment

Stdin (optional — fed to the program on each run)

Expected stdout (exact match after normalization = full execution marks; close match = partial via similarity)

Compilation max marks

Execution max marks

Connect the grader server to enable all phases.

Running… 0%

No compile report yet Click Run compile — gcc only; executable name matches the .c stem.

No rubric report yet Click Run rubric scoring (requires rubric.json on the server).