Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@TutorFlow MCP Show me Aiden's progress and log a note from today's session."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
An MCP (Model Context Protocol) server that exposes a tutoring workflow — students, session notes, homework, progress — as tools an AI agent (Claude Desktop, Cursor, Claude Code) can call directly. Instead of opening a dashboard to check "how is Aiden doing," you can just ask.
Tutors juggle student-specific context (what was covered, what's still shaky, what's due) across notebooks, texts, and memory. This project puts that context behind a small set of well-defined tools so an AI assistant can answer questions and log updates on your behalf, without you re-explaining context every session.
Related MCP server: Student Progress Tracker MCP Server
Five tools implemented and smoke-tested end to end (write tools and the
report tool's error path all verified with a scripted MCP client; the
report tool's actual LLM call needs your own ANTHROPIC_API_KEY to try live):
list_students — list all students, optionally filtered by name/subject
get_student_progress — recent session notes + open homework for one student
log_session_note — write. Defaults to preview-only; pass confirm: true to save.
assign_homework — write. Same preview/confirm: true pattern.
generate_progress_report — the one tool that reasons rather than just
fetches: calls the Anthropic API to turn shorthand session notes into a
short, parent-friendly summary. Defaults to draft-only; pass save: true
to store it in progress_notes. Requires ANTHROPIC_API_KEY.
Streamable HTTP transport is also implemented (src/mcp/http-server.ts) —
tested end to end with a real HTTP client: unauthenticated requests
correctly rejected, session creation, tools/list, tools/call, and
session termination all verified.
Planned next (see build plan):
Angular dashboard as a second frontend onto the same data
Claude Desktop (local) Claude.ai / remote client
│ stdio │ Streamable HTTP
▼ ▼
src/mcp/server.ts src/mcp/http-server.ts
(spawned subprocess) (Express + session map + bearer auth)
└───────────────┬───────────────┘
▼
src/mcp/createServer.ts
— builds a fresh McpServer, registers all 5 tools
│
▼
src/mcp/tools/*.ts — one file per tool: zod schema + handler
│
▼
src/db/client.ts — SQLite connection (better-sqlite3)
tutorflow.db — local dev databaseBoth entry points share the same tool implementations via
createTutorFlowServer() — the tools themselves don't know or care whether
they're being called over stdio or HTTP.
Each tool file follows the same shape on purpose: a zod schema for
arguments, a handler that queries the DB, and a registerX(server) function
called once from server.ts. Adding tool #3 means adding one file and one
line in server.ts — nothing else changes.
npm install
npm run db:init # creates tutorflow.db and seeds 2 sample students
npm run dev # starts the MCP server on stdioAdd this to your Claude Desktop MCP config
(~/Library/Application Support/Claude/claude_desktop_config.json on macOS,
%APPDATA%\Claude\claude_desktop_config.json on Windows):
{
"mcpServers": {
"tutorflow": {
"command": "npx",
"args": ["tsx", "/absolute/path/to/tutorflow-mcp/src/mcp/server.ts"]
}
}
}Restart Claude Desktop, then try asking: "List my students" or "How is Aiden doing?"
# .env needs DB_PATH, ANTHROPIC_API_KEY (optional), and MCP_AUTH_TOKEN (recommended)
npm run dev:httpVisit http://localhost:3000/health to confirm it's up. If MCP_AUTH_TOKEN
is set, every /mcp request needs Authorization: Bearer <token> or it's
rejected with 401 — this is a single shared secret, fine for a solo demo,
not real per-user auth.
Push this repo (see .gitignore — node_modules, dist, and *.db
are already excluded).
On Render or Railway, create a new web service pointing at the repo with
build command npm install && npm run build and start command
npm run start:http.
Set environment variables on the platform: ANTHROPIC_API_KEY,
MCP_AUTH_TOKEN (generate a random string — this is your server's
password, keep it secret), and optionally DB_PATH if you want the
SQLite file somewhere specific. PORT is usually set automatically by
the platform.
Note SQLite lives on local disk — fine for a demo, but if the platform's
filesystem isn't persistent across deploys/restarts your data resets.
For anything beyond a portfolio demo, swap better-sqlite3 for a
Postgres client (pg) pointed at a free Supabase/Neon instance instead —
the query shapes in src/mcp/tools/*.ts stay almost identical.
Two ways, depending on what your Claude plan/client supports:
A. Native custom connector (Claude.ai/Desktop, Pro plan or above) —
Settings → Connectors → Add custom connector → paste your deployed URL
(https://your-app.onrender.com/mcp). If your account has the "Request
headers" option under Advanced settings, add
Authorization: Bearer <your MCP_AUTH_TOKEN> there. If that option isn't
available yet, use option B instead.
B. mcp-remote bridge (works today, any account) — add this to
claude_desktop_config.json instead of a local command/args block.
Windows note: pass the header via an env var to avoid Claude Desktop
mangling the spaces in the value:
{
"mcpServers": {
"tutorflow": {
"command": "npx",
"args": [
"mcp-remote",
"https://your-app.onrender.com/mcp",
"--header",
"Authorization:${AUTH_HEADER}"
],
"env": { "AUTH_HEADER": "Bearer your-mcp-auth-token" }
}
}
}log_session_note, assign_homework, and generate_progress_report (when
saving) all default to a preview-only response and only write when
called again with confirm: true (or save: true for the report). This
means an agent can't silently write to your data on a single ambiguous
request — it has to show you exactly what it's about to save first. Worth
keeping this pattern for every future write tool.
Two ways to provide it, depending on how you're running the server:
Local dev (npm run dev) — copy .env.example to .env and fill in
the key. Loaded automatically via dotenv/config.
Via Claude Desktop — add it to the server's config block instead, so it's set before the process starts:
{
"mcpServers": {
"tutorflow": {
"command": "npx.cmd",
"args": ["tsx", "D:\\tutorflow-mcp\\src\\mcp\\server.ts"],
"cwd": "D:\\tutorflow-mcp",
"env": { "ANTHROPIC_API_KEY": "sk-ant-..." }
}
}
}Assign a homework item to a student. By default this only PREVIEWS the assignment and does not save it — call again with confirm: true to actually create it. Use list_students first if you don't know the student_id.
| Name | Required | Description | Default |
|---|---|---|---|
| confirm | No | Set true to actually create the assignment. Defaults to false (preview only). | |
| due_date | No | Due date, e.g. 2026-09-12 (YYYY-MM-DD). Omit if there's no fixed due date. | |
| student_id | Yes | The student's id, from list_students | |
| description | Yes | What the homework is |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full behavioral burden. It discloses the critical default side-effect (preview only, no save) and the practical workflow of 'call again with confirm: true', deliberately preventing accidental writes. It omits details like reversibility, permissions, or preview output content, but the core safety behavior is well covered.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is three concise sentences, with the purpose front-loaded and no filler. Each sentence earns its place: purpose, behavioral/workflow warning, and prerequisite.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a tool with no annotations and no output schema, the description covers the essential context: action, default side-effect (preview), confirmation requirement, and prerequisite. The only major gap is the content or format of the preview response, which is not specified, but for a straightforward create-with-preview tool the coverage is nearly complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the baseline is 3. The tool description restates the confirm workflow and the list_students dependency, both of which are already present in the schema's parameter descriptions, so it adds little meaning beyond what the structured schema already conveys.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The first sentence uses a specific verb ('assign'), concrete resources ('homework item', 'student'), and is clearly distinct from list_students, get_student_progress, log_session_note, and generate_progress_report. It names the exact operation and target, leaving no ambiguity.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides an explicit precondition: 'Use list_students first if you don't know the student_id.' It also communicates the intended two-step preview-and-confirm flow. It lacks explicit exclusion language for sibling tools, but the purpose differentiation makes those exclusions unnecessary.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Generate a parent-friendly progress summary for a student by having an LLM reason over their recent session notes and homework. Returns a draft by default — call again with save: true to store it in progress_notes. Requires ANTHROPIC_API_KEY.
| Name | Required | Description | Default |
|---|---|---|---|
| save | No | Set true to store the generated report. Defaults to false (draft only). | |
| student_id | Yes | The student's id, from list_students |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations provided, so the description must disclose behavior. It does: returns a draft by default, requires save: true to persist, and requires ANTHROPIC_API_KEY. It clearly discloses the side effect of saving, which is strong disclosure for a tool with no annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two concise sentences that front-load the main purpose and then detail the save behavior and API key requirement. No fluff or unnecessary words.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a tool with a simple schema and no output schema, the description covers the essential operational details: purpose, default behavior, save option, and API key dependency. It could mention the output format but that is not required without an output schema.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema covers both parameters fully (100% coverage). The description adds no new parameter semantics beyond what the schema already states, so the baseline of 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool generates a parent-friendly progress summary by reasoning over session notes and homework. It implicitly distinguishes itself from get_student_progress (summary vs raw data) but does not explicitly name the alternative, so a 4 is appropriate.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No explicit guidance on when to use this tool versus siblings like get_student_progress. The description implies it is for summaries but does not contrast with alternatives or state prerequisites beyond the API key, leaving the agent to infer the use case.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Get a student's recent session notes and open homework by student id. Use list_students first if you don't know the id.
| Name | Required | Description | Default |
|---|---|---|---|
| student_id | Yes | The student's id, from list_students |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of behavioral disclosure. The verb 'get' implies a read-only operation and the returned data is named, which is honest and non-contradictory. But it does not define what 'recent' or 'open' mean, describe the return shape, or note behavior on an invalid/unknown student_id. Adequate but thin for an operation with zero annotation coverage.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences, zero filler. The purpose sentence is front-loaded with verb, resource, and key; the second sentence adds the prerequisite. Every word earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a single-parameter retrieval tool with no output schema, this is nearly complete: it names what is returned, the required input, and the prerequisite. Minor gaps remain — no return-structure description (which a simple tool can tolerate) and no definition of 'recent'/'open' timeframes — but nothing critical blocks correct invocation.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% — the schema already documents student_id as 'The student's id, from list_students', including its source. The description's 'by student id' and 'Use list_students first' only reinforce what the schema already states, adding no new semantic meaning. Baseline 3 is correct when the schema does the heavy lifting.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb ('Get') plus a concrete resource ('a student's recent session notes and open homework') keyed by student id. The scope is unambiguous and easily distinguished from siblings: it retrieves existing data, unlike log_session_note and assign_homework (writes) and generate_progress_report (report generation).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly routes the agent to the prerequisite: 'Use list_students first if you don't know the id.' This gives clear context for calling the tool correctly. It does not, however, mention when to prefer generate_progress_report or log_session_note instead, so some alternative routing is left to inference.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List all students being tutored, with their grade and subject focus. Optionally filter by a search term matched against name or subject.
| Name | Required | Description | Default |
|---|---|---|---|
| search | No | Optional text to match against student name or subject_focus |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of disclosing behavior. It clearly implies a read-only operation (listing) and describes the filter behavior. It does not mention any side effects or return format, but for a simple list tool this is largely implicit. No contradictions exist.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences with zero redundancy. The core purpose is stated first, followed by the filter option. Every word earns its place, making it easy for an agent to parse quickly.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the simplicity of the tool (one optional parameter, no output schema, no nested objects), the description is complete enough for an agent to invoke it correctly. It might benefit from noting the return format (e.g., an array of students), but that is easily inferred from the purpose. Minor gaps like ordering or pagination are not critical for this straightforward tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema already provides 100% coverage for the single optional parameter, describing it as matching against student name or subject_focus. The description adds no new semantic detail beyond restating that the search matches name or subject. Since schema coverage is complete, the baseline of 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the action (List), the resource (students), and the included details (grade and subject focus). It also mentions the optional filter. This differentiates it from sibling tools like get_student_progress or assign_homework, which are clearly not listing operations.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explains when to use the optional filter (search term) but does not explicitly state when to choose this tool over alternatives. However, the sibling names are distinct enough that an agent can infer list_students is for enumerating students, making the usage context clear even without explicit exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Log a note for a tutoring session. By default this only PREVIEWS the note and does not save it — call again with confirm: true to actually write it. Use list_students first if you don't know the student_id.
| Name | Required | Description | Default |
|---|---|---|---|
| note | Yes | What happened in the session | |
| confirm | No | Set true to actually save the note. Defaults to false (preview only). | |
| student_id | Yes | The student's id, from list_students | |
| session_date | Yes | Date of the session, e.g. 2026-09-06 (YYYY-MM-DD) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden. It clearly discloses that the tool only previews by default and requires a second call with confirm: true to actually save, which is a critical behavioral trait. It doesn't describe the preview's return format, but the core behavior is transparent.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences with zero waste. The purpose is front-loaded, followed by the essential preview/confirm behavior, then the prerequisite hint. Every sentence earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a 4-parameter tool with full schema coverage and no output schema, the description covers the key behavioral nuance (preview vs save) and the prerequisite. Nothing an agent needs to call it correctly is missing.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so baseline is 3. The description adds the hint about using list_students for student_id, which is beyond the schema. However, the confirm and session_date details are already in the schema, so the description adds only marginal value.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb 'Log' and resource 'a note for a tutoring session', clearly distinguishing it from sibling tools like list_students or assign_homework. The preview/confirm behavior is mentioned up front, making the tool's function unambiguous.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly instructs to use list_students first if student_id is unknown, providing a clear prerequisite and routing to a sibling. Doesn't explicitly state when not to use this tool, but the guidance is sufficient for an agent to decide.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections. Dates show when Glama detected each change.
v0.1.0assign_homeworkgenerate_progress_reportget_student_progresslist_studentslog_session_noteEach tool targets a distinct action and resource: listing students, retrieving progress, logging session notes, assigning homework, and generating reports. No two tools could be confused for the same purpose.
All tool names follow a consistent verb_noun pattern (list_students, get_student_progress, log_session_note, assign_homework, generate_progress_report). The naming is uniform and predictable.
With 5 tools, the server is well-scoped for a tutoring workflow. Each tool covers a core operation without redundancy or bloat, fitting the ideal 3-15 range.
The tool surface covers the main lifecycle of tutoring: viewing students, accessing progress, adding session notes and homework, and producing reports. Minor gaps exist (no explicit update/delete for notes or homework), but agents can work around these by logging new notes or assignments.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Turn any recording or document into notes, flashcards and quizzes your agent can read and act on
Give your AI agents the tools to build, manage, and run automation workflows.
AI-native task management: list, create, update and archive tasks with rich context for AI agents
Manage tasks, Focus Zone, notes, projects, and task history from compatible AI assistants.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/Nishitadoval/tutorflow-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
