Command Palette

Search for a command to run...

UnylyUnyly
Browse all

Aoa Course Connector

FreeNot checked

Permission-aware AoA connector for authorized course-platform content: normalized knowledge, retrieval, evidence packets, local evals, and MCP tools.

GitHubEmbed

About

Permission-aware AoA connector for authorized course-platform content: normalized knowledge, retrieval, evidence packets, local evals, and MCP tools.

README

aoa-course-connector turns course platforms that the operator is legitimately allowed to access into normalized, source-backed course knowledge for agents. It owns connector method, adapters, schemas, local retrieval logic, evidence packets, an independently runnable MCP server, local evals, and bounded owner-local statistics. It is not a repository for course corpora.

Owner boundary

The repository owns:

  • source policy and adapter contracts;
  • GetCourse and Skillspace browser-session adapters;
  • the Stepik clean API adapter;
  • platform-neutral course, module, lesson, step, asset, transcript, assignment, discussion, topic, entity, and evidence objects;
  • authorized discovery, ingestion, normalization, refresh, and continuity;
  • local keyword and semantic indexes, course graphs, queries, and answers;
  • source-backed CLI and MCP read surfaces;
  • fixture-safe local evals and calibration intake;
  • the root stats/ port over connector-authored evidence.

Runtime deployment belongs to abyss-stack. Shared eval doctrine and proof authority belong to aoa-evals. Shared statistical grammar and cross-owner composition belong to aoa-stats. Private course truth remains in the operator's configured storage.

Source posture

GetCourse and Skillspace use one platform-neutral browser-session pipeline with hard-adapter parsing differences. Stepik demonstrates the clean API path from course to sections, units, lessons, and steps. Moodle and Canvas are reference targets; Coursera, Teachable, Thinkific, and Kajabi remain future adapter routes.

Every route preserves source identity, access mode, capture or update time, freshness, authority, and evidence. Fixture and example sources prove method only. They are never presented as operator-connected live sources.

Data flow

  1. Discovery identifies course entrypoints and may register local sources.
  2. Ingestion captures an authorized or public source into external local storage.
  3. Normalization produces platform-neutral course objects and evidence.
  4. Index and graph builders create rebuildable derived artifacts.
  5. Query surfaces return ranked results, lesson context, answer packets, and evidence chains.
  6. Refresh and continuity surfaces preserve previous artifacts and distinguish source change from bounded or incomplete ingest.

Raw captures, private normalized content, indexes, vectors, graphs, browser state, credentials, and runtime calibration packets stay outside Git under the configured AOA_COURSE_* roots.

Retrieval and evidence

Keyword retrieval uses a versioned BM25 contract. Semantic retrieval supports a deterministic local baseline and an operator-configured HTTP embedding adapter. Hybrid ranking preserves lexical, semantic, freshness, authority, and native course-path contributions instead of hiding them in one opaque score.

Answer packets retain source URL, source id, fetched time, freshness state, authority tier, rank features, refresh hints, and an evidence chain. Cross-source answers keep per-source evidence rather than promoting a blended response into source truth.

Coverage, integrity, and readiness

Ingest coverage distinguishes complete, bounded, partial, and indeterminate source inventories. Browser coverage records the visible lesson population before limits; Stepik coverage compares referenced and fetched structural ids. Identity continuity keeps removal under incomplete ingest inconclusive.

Corpus integrity independently compares normalized canonical objects with keyword and semantic indexes, graph nodes and edges, vectors, postings, and evidence. It is an executable audit, not a replacement for the source.

Readiness is split between local operational readiness and connected-live readiness. A blocked live source does not erase a query-ready local source. Network-touching work requires an explicit gate, and secret values never enter packets or logs.

Connection profiles

aoa_course_connection_profile_v1 is a runtime artifact that carries source refs, local auth-state paths, selected scope, and semantic-provider settings without token values. connection_profile_inspect and connection_profile_status expose redacted inspection and the aoa_course_connection_profile_status_v1 readiness summary, including ready_for_connected_run and network_touched posture. Applying a profile changes only the local source registry.

Local stats port

The root stats/ port currently measures one public reference question: what fraction of structural course-object references declared by the canonical GetCourse, Skillspace, and Stepik starter fixtures are materialized by their adapters. The reference census is 9 / 9 at its named source revision.

This ratio does not prove content adequacy, live-source coverage, corpus integrity, retrieval or answer quality, connector readiness, eval success, or runtime health. Bounded or non-exhausted source inventories are unknown rather than partial success.

MCP and eval boundaries

The aoa-course-connector-mcp package exposes read-oriented source, retrieval, evidence, readiness, calibration, and planning tools over JSON-RPC stdio. It does not own runtime deployment or source truth.

Local eval suites exercise adapter and retrieval behavior. Their results remain local evidence; aoa-evals retains scoring, promotion, verdict, and proof doctrine. Calibration intake creates bounded repair and eval-intake candidates without moving private payloads into Git.

Executable routes

Exact commands are owned by the CLI parser, scripts/validate_connector.py, scripts/verify_agent_install_route.py, tests, and the CI workflow. The root AGENTS.md contains the bounded operator route. Ordinary Markdown documents architecture and contracts without duplicating command catalogs.

Further reading

  • BOUNDARIES.md — public, local, runtime, eval, and stats ownership.
  • docs/ARCHITECTURE.md — component topology.
  • docs/STATUS.md — current durable capability state.
  • docs/CLI_USAGE.md — CLI group and packet semantics.
  • docs/MCP_USAGE.md — MCP tools and authority.
  • docs/AGENT_INSTALL_ROUTE.md — fresh-install states and proof boundary.
  • stats/README.md — local measurement definition and reference posture.

from github.com/8Dionysus/aoa-course-connector

Installing Aoa Course Connector

This server has no published package — it is built from source. Open the repository and follow its README.

▸ github.com/8Dionysus/aoa-course-connector

FAQ

Is Aoa Course Connector MCP free?

Yes, Aoa Course Connector MCP is free — one-click install via Unyly at no cost.

Does Aoa Course Connector need an API key?

No, Aoa Course Connector runs without API keys or environment variables.

Is Aoa Course Connector hosted or self-hosted?

Self-hosted: the server runs locally on your machine via the install command above.

How do I install Aoa Course Connector in Claude Desktop, Claude Code or Cursor?

Open Aoa Course Connector on unyly.org, pick your client tab (Claude Desktop, Claude Code, Cursor) and press Install — the config is generated automatically, no JSON editing.

Related MCPs

Compare Aoa Course Connector with

Not sure what to pick?

Find your stack in 60 seconds

Author?

Embed badge for your README

Browse similar

All browse MCPs