Output Structure¶
After running the pipeline, your output directory (default ./insights) contains:
insights/
├── intake/ # Daily reading digests
│ └── 2026-03-04.md
│
├── journal/ # Dev journal entries
│ └── 2026-03-04.md
│
├── blog/ # Blog posts
│ ├── weekly-2026-W09.md
│ └── multi-agent-patterns.md # Thematic posts
│
├── sessions/ # Per-session Obsidian notes
│ └── session-note.md
│
├── daily/ # Daily session summaries
│ └── daily-2026-03-04.md
│
├── weekly/ # Weekly session digests
│ └── week-2026-W09.md
│
├── projects/ # Per-project notes
│ └── my-project.md
│
├── brainstorm/ # Content ideas
│ └── 2026-03-04.json
│
├── index.md # Session index
│
# State files (hidden, managed automatically):
├── .distill-memory.json # Unified cross-pipeline memory
├── .distill-seeds.json # Seed ideas
├── .distill-shares.json # Shared URLs
├── .distill-notes.json # Editorial notes
├── .distill-content-store.json # Vector store (JSON fallback)
├── .distill-graph.json # Knowledge graph
├── .intake-state.json # Intake deduplication
├── .distill-last-run.json # Last pipeline run report
├── .daily-social-state.json # Social post state
├── .working-memory.json # Journal memory
├── .intake-memory.json # Intake memory
├── .blog-state.json # Blog generation state
└── .blog-memory.json # Blog cross-post memory
Obsidian integration¶
The default publisher uses Obsidian-compatible markdown:
- YAML frontmatter with tags, dates, and metadata
[[Wiki links]]for cross-referencing- Mermaid code blocks for diagrams
Drop the insights/ folder into an Obsidian vault and everything links together.
State files¶
State files (prefixed with .) are managed automatically. They track:
- Which articles have been processed (deduplication)
- Cross-session memory (themes, entities, trends)
- Blog generation state (which posts exist)
- Seeds and shares (pending vs used)
You can safely delete any state file to reset that subsystem. The pipeline will rebuild from scratch on the next run.