Last updated: July 19, 2026
- We open-sourced
learning-orchestrator-mcpv1.0.0, a stateless MCP server that structures progressive curricula via Bloom's Taxonomy, schedules review using the SuperMemo SM-2 spaced-repetition algorithm, and syncs progress to external LMS platforms. - It's the third release in SEOSiri's MCP ecosystem, alongside the Bio-Robotics Core Engine and the API Guard.
- Confirmed live via MCP Inspector: the server exposes 5 tools —
get_progressive_syllabus,generate_skills_assessment,calculate_spaced_repetition,sync_ai_platform_state, andsync_lms_onboarding_state— with a real tool call verified end to end. - Its demo curriculum segments mirror its two sibling projects — robotics/kinematics, API security/compliance, and SEO/AEO — so it can teach and test the exact skills those other tools require.
What Does the Learning Orchestrator Actually Do?
It's a stateless MCP server that structures learning into progressive stages using Bloom's Taxonomy, schedules review of previously learned material using the SuperMemo SM-2 spaced-repetition algorithm, and is designed to sync assessment results back to external Learning Management Systems. It's built to let an AI tutoring agent or an LMS hand off syllabus structure, review scheduling, and progress tracking to a dedicated, deterministic layer instead of relying on an LLM to improvise instructional design decisions on the fly.
Spaced repetition and staged learning objectives are both well-established, independently studied approaches in education — the difficulty has historically been operational: wiring them into an actual LMS, automating the scheduling math, and keeping the connection between platforms secure. learning-orchestrator-mcp targets that operational gap, using the Model Context Protocol (MCP) as the connection standard.
Why This Is the Third Piece of a Larger Pattern
This is SEOSiri's third open-source MCP server in quick succession, and it's worth naming the pattern directly: each one takes a domain that already has established standards — robotics kinematics and G-code, OWASP/PCI/HIPAA/GDPR compliance, and now educational theory like Bloom's Taxonomy and SM-2 spaced repetition — and builds a stateless MCP translation layer that lets an AI agent act on that domain safely and consistently, rather than inventing new theory. The Learning Orchestrator's documented subject segments make this explicit: its curriculum categories are Robotics & Kinematics, API Security & Compliance, and Digital Marketing & SEO/AEO — the same three domains covered by its two sibling projects. Read together, the three tools form a small ecosystem: build the hardware/security/marketing tooling, then teach people, or onboard new team members, on how to use it, through the same protocol.
| Server | Domain | What It Does | Live-Tested? | |
|---|---|---|---|---|
| Bio-Robotics Core Engine | Robotics & kinematics | Translates biological data and EMG signals into G-code for physical actuation | Yes — 6 tools, verified live outputs | |
| API Guard | API & hardware-command security | Sanitizes and remediates LLM-generated payloads against OWASP/PCI/HIPAA/GDPR/biorobotics limits | Yes — 1 tool, verified live output | |
| Learning Orchestrator (this post) | EdTech / skills development | Structures progressive syllabi, schedules SM-2 spaced repetition, syncs to LMS | Yes — 5 tools, verified live output |
What the Curriculum Covers
Per the project's documentation, three subject segments are defined so far:
- Robotics & Kinematics: Cartesian coordinate space mapping, G-code instructions, and deck calibration — the same concepts underpinning the Bio-Robotics Core Engine.
- API Security & Compliance: OWASP injection patterns, HIPAA PII/PHI validation concepts, and cryptographic signing concepts — the same territory covered by the API Guard.
- Digital Marketing & SEO/AEO: JSON-LD schema design, GDPR privacy structures, and conversational voice-search optimization.
These read as demonstration curricula built from SEOSiri's own existing technical domains rather than a general-purpose subject bank — a reasonable starting point for a v1.0.0 release, and consistent with the project's stated openness to custom subject-segment mapping for other domains.
The Learning Science Behind It
- Bloom's Taxonomy structures learning objectives into a hierarchy — from basic recall through analysis and evaluation — used here to stage progressive syllabi rather than presenting all material at a single difficulty level.
- SuperMemo SM-2 is a specific, published spaced-repetition algorithm (not a generic description) that schedules the next review date based on a "quality of recall" grade and an evolving ease factor. It's the same family of algorithm behind tools like Anki, applied here to automated skill tracking instead of manual flashcards.
Live-Tested: The Full 5-Tool Suite
Confirmed directly via Glama's browser-based MCP Inspector: the server responds to a tools/list request with five tools, and a real call to sync_lms_onboarding_state was executed and returned a live result.
sync_lms_onboarding_state mapping a 95% SCORM quiz score to an SM-2 grade, scheduling the next review, and generating a signed LTI Outcomes payload — with all five tools visible in the Inspector's tool list.| Tool | Description |
|---|---|
get_progressive_syllabus |
Retrieves the exact deterministic lesson plan based on the learner's current mastery level. |
generate_skills_assessment |
Generates targeted questions mapped strictly to a Bloom's Taxonomy level. |
calculate_spaced_repetition |
Calculates the next optimal review date using the standard SM-2 algorithm. |
sync_ai_platform_state |
Verifies and synchronizes student progress state with a connected AI platform. |
sync_lms_onboarding_state |
Integrates with Moodle/Canvas via LTI, mapping quiz scores to SM-2 metrics and signing an LTI Outcomes payload. |
The live test called sync_lms_onboarding_state with lms_student_id="moodle_student_88", scorm_package_id="scorm_compliance_01", quiz_score_pct=95.0, repetition_count=3, previous_ease_factor=2.5, and previous_interval_days=6. The server returned:
- SM-2 metrics:
next_interval_days: 15,updated_ease_factor: 2.6,next_repetition: 4— a 95% score correctly mapped to the maximum SM-2 grade (5), which raises the ease factor by exactly 0.1 (2.5 → 2.6) under the standard formula, and 6 × 2.6 = 15.6 truncates to the reported 15-day interval. Independently checked against the published SM-2 algorithm and exact. - LTI Outcomes payload: a realistic
lis_outcome_service_urlpointing at a Moodle-stylemod/lti/service.phpendpoint, alis_result_sourcedidcombining the student and package IDs, and a 64-character hexlti_handshake_signature— the correct length for a SHA-256-based HMAC output. - Status:
LMS_SYNC_SUCCESS.
That combination — correct SM-2 math plus a correctly-shaped cryptographic signature plus a realistic LTI endpoint pattern — is meaningfully harder to fake by coincidence than a single plausible-looking field, which is why this section is now presented as verified rather than reported.
Secure LMS & Cross-Platform Handshakes
The verified sync_lms_onboarding_state tool above already does the core of this: it maps a raw SCORM quiz score into SM-2 metrics, schedules the next review, and compiles an LTI Outcomes payload signed with HMAC-SHA256 to synchronize state back to a host LMS like Moodle or Canvas. Worth being precise about what that signature does and doesn't establish: HMAC signing lets the receiving LMS verify a payload wasn't altered in transit and came from a holder of the shared secret — which is meaningfully different from an absolute guarantee against all tampering or exfiltration risk. Custom, deeper enterprise LMS integration work beyond this shipped tool is described separately as B2B consulting, not something bundled into the open-source v1.0.0 package by default.
Worth noting for anyone evaluating actual LMS compatibility: the documentation doesn't specify whether these handshakes target xAPI (the modern ADL learning-record standard), SCORM (the older, still widely deployed ADL packaging standard), or LTI (the standard most common in university and K-12 platforms). Those three are not interchangeable — an LMS built around SCORM won't automatically speak xAPI or LTI — so confirm which standard this project targets before assuming compatibility with your specific platform.
Who This Is For
- Teams building AI tutoring or onboarding agents. If you're already using Claude Code, Cursor, or another MCP client to build educational or training tooling, this gives you progressive syllabus structure and SM-2 review scheduling as a pluggable layer instead of building it in-house.
- Corporate L&D and onboarding teams. The documented use case — mapping proprietary technical manuals into automated, spaced-repetition assessment banks — is a direct fit for structured employee onboarding, particularly for technical roles.
- Developers already using SEOSiri's other MCP servers. Since the demo curricula cover robotics/kinematics and API security concepts directly, it's a plausible way to onboard a new team member onto the Bio-Robotics Core Engine or API Guard specifically.
Quickstart
# Install locally in editable development mode
pip install -e .
pytest tests/test_learning.py
# Or run directly from GitHub with zero setup (requires uv)
# — add this to your claude_desktop_config.json:
{
"mcpServers": {
"seosiri-learning-orchestrator": {
"command": "uv",
"args": ["run", "--github", "SEOSiri-Official/learning-orchestrator-mcp", "src/main_server.py"]
}
}
}
Technical Specifications
- Communication protocol: Model Context Protocol JSON-RPC over stdio.
- Verification: automated unit testing via pytest (
tests/test_learning.py). - Containerization: a Dockerfile is included for isolated runtime environments.
- Directory sync: a
glama.jsonfile drives the automatic Glama.ai listing sync. - Release: tagged
v1.0.0, "Initial Production Release," July 18, 2026 — the only release currently published. - License: MIT, per the project's own LICENSE file (note: the README's LICENSE link currently points to a different SEOSiri repository and should be corrected at the source).
Listed on the Glama.ai MCP Directory
The project is listed on Glama.ai's MCP directory under Developer Tools, Education & Learning Tools, and Knowledge & Memory. The source is available on GitHub, currently at 0 stars and 0 forks — a brand-new release, reported here as-is rather than dressed up.
Voice Search & Conversational Query Answers
What is the SEOSiri Learning Orchestrator? An open-source, stateless MCP server that structures progressive learning syllabi using Bloom's Taxonomy and schedules review using the SuperMemo SM-2 spaced-repetition algorithm, with LMS integration.
Is this the same as the Bio-Robotics Core Engine or the API Guard? No, it's a separate, third MCP server in the same SEOSiri ecosystem, but its demo curriculum covers the technical concepts behind both of those tools.
What is spaced repetition? A learning technique that schedules review of material at increasing intervals timed to counteract natural forgetting. SM-2, the specific algorithm this server uses, calculates each new interval from a recall-quality grade and an evolving ease factor.
What is Bloom's Taxonomy? A hierarchy of learning objectives, from basic recall through analysis and evaluation, commonly used to structure curricula so learners build foundational knowledge before tackling higher-order skills.
Does it come with pre-built courses? The documented release includes three demonstration subject segments — robotics/kinematics, API security/compliance, and SEO/AEO — and the project is described as open to custom subject-segment mapping for other domains.
Has the tool suite been independently verified live? Yes. A live MCP Inspector session confirmed all 5 tools via a tools/list call, and a real execution of sync_lms_onboarding_state returned output matching the published SM-2 algorithm exactly, plus a correctly-shaped HMAC-style signature and a realistic LTI endpoint.
Does it work with SCORM, xAPI, or LTI-based LMS platforms? The project's documentation doesn't specify which standard its LMS handshakes target, and SCORM, xAPI, and LTI are not interchangeable — confirm compatibility with your specific platform before assuming it will connect out of the box.
Is it open source? Yes, under the MIT License, with the source available on GitHub.
Reference Sources
- Model Context Protocol Specification — modelcontextprotocol.io
- Vanderbilt University Center for Teaching — Bloom's Taxonomy
- IETF — RFC 2104: HMAC
- Advanced Distributed Learning Initiative — xAPI
- 1EdTech — Learning Tools Interoperability (LTI)
Explore or Contribute to the Project
The project is fully open-source. Explore the codebase, file an issue, or get involved in development directly on GitHub.
View on GitHub ➔ Sponsor the Project
