Skip to content

Document the universal loop #2

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
axAilotl wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Document the universal loop #2

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 11 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,6 +36,17 @@ Topic-scoped compilation over selected source folders, types, tags, and terms.
3. Analyst / Agent
Higher-cost interactive work over the compiled layer for synthesis, comparison, and refinement.

### The Universal Loop

The layers are not a one-way pipeline — they form a loop that compounds:

1. **Capture** — bookmarks, papers, clippings, and transcripts flow in raw.
2. **Compile** — processors and the archivist turn them into cited wiki pages.
3. **Query** — you and agents read the compiled layer.
4. **Feed back** — `wiki-query --write-back` persists curated query results as new wiki pages, the archivist recompiles topics when their input manifest changes, and digests resurface unread material — which drives the next round of capture.

Each pass makes the wiki a better source for the next one, without raw material ever being rewritten.

## Quick Start

### Setup
Expand Down
11 changes: 11 additions & 0 deletions docs/project_overview.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,6 +29,17 @@ The problem it solves is not capture — it's that captured knowledge rots. Book
2. **Processing** — `core/ingestion_runtime.py` dispatches queued artifacts to `processors/`: tweet/thread enrichment, document processing (papers, PDFs, READMEs), transcription (Whisper/Deepgram), and LLM tasks (tags, summaries, alt-text, translation) via a routed multi-provider `LLMInterface` with caching.
3. **Archivist / Analyst** — Topic-scoped compilation: `archivist_retrieval/` selects sources via full-text + pgvector semantic search, the compiler synthesizes cited wiki pages, and agents query the result through `AgentSurfaceService` (API + MCP).

## The Universal Loop

The layers are not a one-way pipeline — they form a loop that compounds:

1. **Capture** — bookmarks, papers, clippings, and transcripts flow in raw.
2. **Compile** — processors and the archivist turn them into cited wiki pages.
3. **Query** — you and agents read the compiled layer.
4. **Feed back** — `wiki-query --write-back` persists curated query results as new wiki pages, the archivist recompiles topics when their input manifest changes (`core/wiki_change_provenance.py`), and digests resurface unread material — which drives the next round of capture.

Each pass makes the wiki a better source for the next one, without raw material ever being rewritten.

## Storage contract (the core design rule)

- `knowledge_vault/` — synced vault: raw + generated artifacts (`tweets/`, `threads/`, `library/`, `pdfs/`, `repos/`, `transcripts/`, `media/`, `_digests/`).
Expand Down