Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
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
2 changes: 1 addition & 1 deletion PLANS.md
Original file line number Diff line number Diff line change
Expand Up @@ -1113,7 +1113,7 @@ invalidation; the 2.27 MB bounded view is built on demand, shared per revision,
and served with an exact ETag. On retained local state, a matching conditional
read returned 304 with zero bytes in 0.000789 seconds, and the observed idle
control-plane sample fell from the old 52–55% projection-fanout workload to
0.2%. Studio retains its last good view and reports Live data, Updating, or
0.2%. Studio retains its last good view and reports SSE connected, Updating, or
Reconnecting instead of silently replacing megabytes of state.

The product bottleneck has moved from discovery volume to opportunity delivery.
Expand Down
18 changes: 11 additions & 7 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -116,16 +116,20 @@ smoke commands, read [Operations](docs/OPERATIONS.md).

## First use

1. Open Studio and check **Readiness** for the control plane, catalog, storage,
Agent runtime, and credential posture.
1. Open Studio. The default **Discover** page reports **System ready** source
health and a **Refresh catalogs** first step. The same refresh also lives on
**System overview**.
2. Refresh anonymous catalogs. A refresh creates a new immutable corpus; it
does not spend model budget or grant trading authority.
3. Open the ontology / discovery workspace and inspect the standing campaigns.
Run a bounded campaign or issue against the retained corpus.
4. Read the effect timeline, exact listing references, counterexamples, and
3. If discovery is blocked (`credential unavailable`), open **Agent operations**
to inspect the existing Codex session. Stay on Discover to keep refreshing
catalogs without DeepSeek or Codex; **Explore next** is not the next click.
4. After catalogs are in view, start a heuristic scan or inspect standing
campaigns against the retained corpus.
5. Read the effect timeline, exact listing references, counterexamples, and
token usage. An empty or falsified run is retained research evidence.
5. Move only a grounded multi-listing hypothesis into independent review.
6. Treat economic hints as routing signals until fresh books, fees, depth, and
6. Move only a grounded multi-listing hypothesis into independent review.
7. Treat economic hints as routing signals until fresh books, fees, depth, and
the exact verifier all agree.

The longer operator walkthrough is in [Studio](docs/STUDIO.md).
Expand Down
Loading