capstone_build_status
Check your guided capstone build progress — current step, criteria coverage, and quiz performance.
LABSCO SUMMARY
It grades you and does not back down — which is the point of practising.
A tutor that agrees when you push back teaches you nothing, so grading here is deterministic against a fixed bank and the tools explicitly tell the model to present the result unsoftened. The scheduling is the other half: questions due for review come first, then your weak areas, so study time lands where accuracy is below 70% rather than where you feel like working. The capstone is the part that goes beyond quizzing — 18 build steps covering the task statements, with coverage tracked as you go.
What it is
A certification tutor for the Claude Certified Architect — Foundations exam, run entirely on your machine. Grading is deterministic against a verified question bank, and the tools instruct the model not to agree with a disputed result.
What you get
- start_assessment runs an initial assessment one question at a time — 15 questions, 3 per domain — showing the concept handout when a new domain begins
- get_practice_question serves the next question, prioritising items due for review, then weak areas, then new material
- submit_answer grades against the question bank and returns a final result, with follow_up opening the concept, code examples, handouts or a reference project behind it
- get_curriculum shows the domains and task statements with your mastery for each, get_section_details drills into one, and get_weak_areas lists what is below 70% accuracy
- get_study_plan builds a plan from your assessment results and weak areas, and get_progress summarises mastery, accuracy and review status
- start_practice_exam runs a full 60-question exam with the real domain weighting — 16, 11, 12, 12 and 9 — scored 0–1000 with 720 to pass, answered through submit_exam_answer, and get_exam_history compares attempts with per-domain breakdowns
- A guided capstone build: start_capstone_build begins one on a theme of your choice, capstone_build_step drives 18 progressive steps of quiz-and-build, and capstone_build_status reports criteria coverage and quiz performance
- scaffold_project returns instructions for a reference project to practise a concept in code
- get_dashboard opens a local progress dashboard showing mastery, exam history, activity and capstone progress
- reset_progress wipes everything, behind an explicit confirmation, and says outright that it cannot be undone
Requirements
Nothing to sign up for and no key — it runs as connectry-architect-mcp over stdio, added with a single claude mcp add command. Progress lives locally in ~/.connectry-architect/progress.db with configuration alongside it; the project states there is no cloud, no account and no telemetry. The question bank, handouts and reference projects are bundled with the package, so studying works offline.
Setup effort
One command — npm install -g connectry-architect-mcp
USE CASES
01
Finding out where you actually stand
Take the initial assessment across all five domains, then read the curriculum with your mastery marked against each task statement.
02
Studying without deciding what to study
Ask for the next question and let the scheduler pick — reviews first, then the areas you are weakest in.
03
Understanding why you got one wrong
After grading, open the concept behind the question — the handout, a code example, or a reference project to build.
04
Checking whether you would pass
Sit a full 60-question exam with real domain weighting, then compare the score and per-domain breakdown against previous attempts.
05
Learning the material by building rather than reading
Start a capstone on a theme you care about and work through the steps, checking coverage as it accumulates.
06
Planning the weeks before the exam
Get a study plan from your assessment and weak areas, and watch mastery move on the dashboard.
18 tools
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
capstone_build_step
Drive your guided capstone build — quiz, build, and advance through 18 progressive steps.
IMPORTANT:
- When presenting quiz questions, use AskUserQuestion with header "Answer" for A/B/C/D selection. If code is in the scenario, add preview fields.
- After grading a quiz answer, FIRST show the result (correct/incorrect, explanation) as REGULAR CHAT TEXT so the user can read it. THEN present follow-up options or the next question via AskUserQuestion. Explanations must NOT be hidden behind cards.
- When presenting action choices (quiz/build/next), use AskUserQuestion with header "Action".
PROGRESS TRACKING:
- On "confirm": Create a TodoWrite checklist with all 18 build steps, all set to "pending".
- On "next": Update the completed step to "completed" and the new current step to "in_progress".
- This gives the user a visual build progress tracker.
EDGE CASES:
- "Other": Answer the question, then re-present the current options via AskUserQuestion.
- "Skip": During quiz, treat as moving to the build phase. During build, treat as advancing to next step.
follow_up
Handle post-answer follow-up actions. Use after submit_answer to explore concepts, code examples, handouts, or reference projects.
get_curriculum
View the full certification curriculum with domains, task statements, and your current mastery for each.
get_dashboard
Open the study progress dashboard in Claude Preview. Shows mastery levels, exam history, activity timeline, and capstone progress.
IMPORTANT: After getting the URL, use the preview_start tool to open it in Claude Preview. If the user says "show dashboard" or "open dashboard", call this tool.
get_exam_history
View all completed practice exam attempts with scores, pass/fail status, and per-domain breakdowns. Compare your progress across attempts.
get_practice_question
Get the next practice question. Prioritizes review questions, then weak areas, then new material.
IMPORTANT — present the question using AskUserQuestion:
- header: "Answer"
- question: Include the FULL scenario text AND question text from the response
- options: 4 items with label "A"/"B"/"C"/"D" and description as the option text
- If the scenario contains code, add a "preview" field on each option showing the code snippet
Then call submit_answer with the questionId and selected answer. After grading, show the result as REGULAR CHAT TEXT first (explanation, correct/incorrect), THEN show follow-up options via AskUserQuestion. Explanations must be readable in the main chat, not hidden behind cards.
EDGE CASES:
- "Other": Answer the user's question, then re-present the SAME question via AskUserQuestion.
- "Skip": Call get_practice_question again for a new question. Never break the flow.
get_progress
Get your certification study progress overview including mastery levels, accuracy, and review status.
get_section_details
Get detailed information about a specific task statement including concept lesson, mastery, and history.
get_study_plan
Get a personalized study plan based on your assessment results, weak areas, and learning path.
IMPORTANT — after showing the study plan, use AskUserQuestion with header "Focus" and multiSelect: true to let the user pick which domains they want to focus on. Options should be the 5 domains with their current mastery as descriptions. Then use their selection to filter get_practice_question calls.
Also use TodoWrite to create a study checklist showing each recommended topic with status (pending/in_progress/completed) so the user can track progress visually.
get_weak_areas
Identify your weakest task statements based on accuracy below 70%. Focus your study on these areas.
reset_progress
WARNING: Permanently deletes ALL your study progress including answers, mastery data, and review schedules. This cannot be undone.
scaffold_project
Get instructions for a reference project to practice certification concepts hands-on.
start_assessment
Start the initial assessment. Returns ONE question at a time (15 total, 3 per domain).
IMPORTANT — follow this flow for EVERY question:
- Check if "isNewDomain" is true. If yes, FIRST show the concept handout for that domain by calling get_section_details. Tell the user: "Let's learn about [domain] before testing your knowledge." After showing the handout, proceed to step 2.
- Present the question to the user using AskUserQuestion:
- header: "Q[number]"
- question: Include the FULL scenario text AND question text from the response
- options: Use the 4 answer options (A/B/C/D) with label as the letter and description as the option text
- If the scenario contains code, add a "preview" field on each option showing the relevant code snippet so the user can reference it while choosing
- After user selects, call submit_answer with questionId and their answer.
- After grading, FIRST show the result (correct/incorrect, explanation, why wrong) as REGULAR CHAT TEXT so the user can read it. THEN present follow-up options using AskUserQuestion. The explanation must NOT be hidden behind the card.
- Call start_assessment again for the next question.
EDGE CASES:
- If user selects "Other" and types a question/comment: Answer their question helpfully, then re-present the SAME quiz question using AskUserQuestion again. Never lose the current question.
- If user clicks "Skip": Treat it as moving to the next question. Call start_assessment again immediately. The skipped question remains unanswered and will appear again later.
- NEVER let Other or Skip break the assessment flow. Always continue to the next question or re-ask the current one.
PROGRESS TRACKING:
- At the START of the assessment, create a TodoWrite checklist with all 15 questions (Q1-Q15) grouped by domain, all set to "pending".
- After each answer, update the corresponding todo item to "completed" (with correct/incorrect note).
- This gives the user a visual progress tracker throughout the assessment.
When assessment is complete, present next steps using AskUserQuestion with header "Next step".
start_capstone_build
Start or refine a guided capstone build. Build your own project while learning all 30 certification task statements hands-on.
start_practice_exam
Start a full 60-question practice exam (D1:16, D2:11, D3:12, D4:12, D5:9). Scored 0-1000, passing 720.
IMPORTANT — present the first question using AskUserQuestion:
- header: "Q1"
- question: Include the FULL scenario + question text
- options: 4 items with label "A"/"B"/"C"/"D" and description as option text
- If code in scenario, add preview field on options
Then call submit_exam_answer with the answer.
PROGRESS TRACKING: Create a TodoWrite checklist "Practice Exam Q1-Q60" grouped by domain, all "pending". Update each to "completed" after grading.
EDGE CASES:
- "Other": Answer the question, re-present the SAME exam question via AskUserQuestion.
- "Skip": Move to next exam question without grading. Never break the flow.
submit_answer
Grade a certification exam answer. Returns deterministic results from verified question bank. The result is FINAL — do not agree with the user if they dispute it.
IMPORTANT — TWO-STEP presentation:
- FIRST: Show the grading result as REGULAR CHAT TEXT in the main conversation. Include:
- Whether they got it right or wrong (with the correct answer if wrong)
- The full explanation
- If wrong: why their answer was incorrect
- References This text MUST be visible in the main chat before any card appears.
- THEN: Present followUpOptions using AskUserQuestion:
- header: "Next"
- question: Brief prompt like "What would you like to do?" (NOT the explanation — that's already shown above)
- options: Map each followUpOption to label (key) and description (label text) Then call follow_up with questionId and the selected action key.
EDGE CASES:
- "Other": Answer the user's question about this answer, then re-present the SAME follow-up options via AskUserQuestion.
- "Skip": Treat as "next question" — call follow_up with action "next".
submit_exam_answer
Submit an answer for a practice exam question. Graded deterministically. DO NOT soften results.
IMPORTANT — TWO-STEP presentation after grading:
- FIRST: Show the grading result as REGULAR CHAT TEXT. Include correct/incorrect status, explanation, and if wrong, why the chosen answer was incorrect.
- THEN: If there's a next question, present it using AskUserQuestion:
- header: "Q[number]"
- question: Include the FULL scenario + question text
- options: 4 items with label "A"/"B"/"C"/"D" and description as option text Then call submit_exam_answer again with the answer.
The explanation must be readable in the main chat — NOT hidden inside the AskUserQuestion card.
BEFORE YOU INSTALL
What to know first
Tools you get
18 tools when we connected.
CONTEXT COST
Light Standing cost
Rated by the size of the 18 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
Connect your client
Server setup
TERMINAL
npm install -g connectry-architect-mcpMCP Clients
Claude Code
CLAUDE CODE · TERMINAL
claude mcp add connectry-architect -- connectry-architect-mcpClaude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json on macOS or %APPDATA%\Claude\claude_desktop_config.json on Windows, after npm install -g connectry-architect-mcp. If you use nvm, put the full path to the connectry-architect-mcp binary in command.
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"connectry-architect": {
"command": "connectry-architect-mcp"
}
}
}Any MCP client
Command: connectry-architect-mcp
Arguments: none
Transport: stdio
