Product frame

Org2 should compete on the things plain prose tools leave underspecified: planning semantics, graph health, provenance, publishing, and agent collaboration over local files.

Org2 earns its complexity when it makes a personal knowledge base operational:

• tasks and planning become agenda data

• links, IDs, aliases, and properties become graph data

• raw notes and captured material become curated notes

• curated notes compile into views, reports, lint results, and published sites

• provenance gives generated artifacts clear origins

Compiler model

The intended corpus flow is:

raw/ -> notes/ -> compiled/ -> views/ -> publish/

• raw/ holds inboxes, imports, transcripts, scraped material, and other source inputs.

• notes/ holds the human-edited canonical corpus.

• compiled/ holds generated machine-readable artifacts.

• views/ holds generated human-facing local reports and review surfaces.

• publish/ holds deployable/public artifacts.

Generated zones should be marked, lintable, and reproducible from source notes where possible.

First-class semantics

The compiler layer should keep moving more behavior out of editor-specific conventions and into portable Org2 semantics:

• tasks, TODO states, planning timestamps, priorities, and tags

• IDs, aliases, file links, wiki links, backlinks, and graph edges

• properties, provenance, generated status, and artifact roles

• source ranges that let tools point back to the original file and heading

• health checks that detect broken links, duplicate IDs, stale generated files, and unsafe corpus boundaries

Implementation direction

Near-term work should bias toward small compiler passes and CLI-backed workflows that editor integrations can reuse.

Good examples:

• richer org2 lint checks for corpus structure and generated artifacts

• explicit provenance fields for generated outputs

• query/search commands that include citations back to source files

• safer roam/linkify passes that can preview changes and exclude generated or archived paths

• publishing docs that explain the corpus model and its user-facing workflows

This keeps Org2 useful today while preventing the project from collapsing into editor-specific glue.

Non-goals

• No Emacs runtime dependency.

• No need to win as a generic prose editor.

• No byte-for-byte clone of every Org Mode behavior before shipping useful workflows.

• No opaque generated artifacts that cannot be traced back to source notes.

Done means compiled artifacts matter

A mature Org2 setup should make notes feel like source code for a personal knowledge system: easy to edit, but also compilable into agenda views, graph artifacts, maintenance reports, and published outputs.