back to projects Shiplog mobile dashboard

Shiplog

A material-first narrative memory layer for indie projects. Shiplog turns real product materials — PRs, commits, Linear issues, screenshots, and release notes — into destination-specific publishable content, grounded in evidence rather than vibes.

Next.js + TypeScript + React 19 Supabase 1,507 tests
In active development — private repository
1,507 Tests
6 Pipeline stages
6 Product surfaces

How it works

Shiplog's pipeline is built from explicit, inspectable seams:

  1. Import project materials with provenance — text, links, repos, images, files
  2. Extract durable facts into Project Memory, linked to source materials
  3. Review, accept, or retract facts to curate project understanding
  4. Assemble artifact-specific briefs from selected materials and active facts
  5. Generate destination-specific Artifacts — blog posts, release notes, build logs
  6. Edit, refine, copy, or export publishable output per destination

Key product surfaces

/posts

Material-first publishing workspace for product intros, updates, and build stories.

/materials

Project Memory panel, reviewing and curating extracted facts.

/outputs

Destination-specific outputs for reviewing generated artifacts.

Visual Renderer

Adapter boundary for generated visual exports (PNG cards, previews).

Narrative Engine

Artifact definitions, brief assembly, plan creation, draft generation.

Xiaohongshu Carousel

Destination adapter for multi-slide content generation.

Engineering principles

No fake intelligence — facts must link to source materials through source_material_ids. No silent degradation when data is missing or corrupt. Artifacts carry explicit provenance so consumers can distinguish canonical data from cached fallbacks.