Skip to content

Milestones

List view

  • Ingest: Broader Content and Dataset type support (Images, PDF, text, Projects)

    Obsidian: - Catalog PKM - Single Dataset - Indexing Strategy varies based on ontology - FTS and Hybrid search bounded within dataset, catalog, or within one of its collections - Per document Metadata collection Heptabase, other PKMs: same - Projects: - Catalog Projects - Single Dataset - Indexing varies wildly - FTS and hybrid search bounded within dataset, catalog, or within one of its collections - Corpus-level metadata URLs: - Catalog Web - Optional dataset (Raindrop, safari etc) - Indexing markdown after conversion - Highlighst support - Hierarchical folders needed - Metadata will be associated later

    No due date
    •0/6 issues closed
    0% complete6 open0 closed

  • Agent: Substrate tool use for agents

    No due date
    •0/1 issues closed
    0% complete1 open0 closed

  • Tooling: Reusable components for agentic development

    Normalization of a stack (python, typescript etc) to provide pre-decided elements. Simplifies agent-based development.

    No due date
    •0/9 issues closed
    0% complete9 open0 closed

  • Evals and Improvements

    No due date
    •0/3 issues closed
    0% complete3 open0 closed

  • Ingest: Ontology Alignment

    No due date
    •0/4 issues closed
    0% complete4 open0 closed

  • Search: Hybrid Search

    No due date
    •0/6 issues closed
    0% complete6 open0 closed

  • Ingest: Indexing an Obsidian Vault with consistent frontmatter extraction

    Ability to ingest and index an Obsidian vault, repeatably, using stored configuration. Ingestion can be re-run, allowing me to iterate on metadata, search capabilities etc In-scope: - A persistent Dataset is created (relational) - Vault can be re-indexed manually - Configuration-based - Note frontmatter is converted to document metadata using user-specified structure (not ontology) and persisted - Documents have consistent unique identifiers that aren't machine-specific - FTS indexing - Simple vector search Out of scope: - Linking to an ontology (for both URI assignment, content classification, and relationship extraction) - Intra-document relationships established - Document-aware chunking - Metadata-aware search

    No due date
    •4/4 issues closed
    100% complete0 open4 closed