Your agent,
meet Canvas
Ask your agent about courses, deadlines and feedback.
Open source. Runs on your computer.
Terminal setup
Install with uv, then try the demo in Codex. No Canvas account needed.
Canvas connection guideInstall
uv tool install https://github.com/liltaket/folava-mcp/releases/download/v0.1.4/folava_mcp-0.1.4-py3-none-any.whlConnect the demo to Codex
codex mcp add canvas-demo -- folava-mcp serve --demoPython 3.11+ · Linux, macOS and Windows · v0.1.4
What you can ask
A fictional course. No account is connected to this site.
For scheduled checks, connect the change feed to your own scheduler.
Environmental fieldwork
“What should I prepare for the fieldwork seminar?”
Field observation report
Record three observations and discuss the conditions. Bring your notes to the seminar.
Field guide
Includes source links and Canvas IDs.
“Help me understand the feedback on my report.”
Report feedback
Read available comments and rubric feedback alongside the assignment instructions.
Submission history is included when available.
“Find the field guide attached to this assignment.”
field-guide.txt
Observe carefully. Record time, place and conditions.
Read supported files in your agent, or export them from the terminal.
Data and access
Folava uses reviewed Canvas GET requests. Uploads, submissions and options that mark content as read are blocked. Canvas may still record access or refresh progress data.
Keep credentials outside your chat, in your OS keyring or a protected secret source.
Your MCP host may send course material to its model provider. That provider’s data terms apply.
Data handlingDevelopment preview
Folava is an independent development preview. Instructure approval for this MCP integration has not been obtained. Deployment limits.
Access depends on your account permissions and institution. External timetables, locked content and third-party tools may be unavailable. Check the originals in Canvas.