Skip to content

chore: sync CONTEXT.md with merged PRs #6-#8 #9

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

Merged
Ishaan29 merged 1 commit into from
Jul 15, 2026
Merged

chore: sync CONTEXT.md with merged PRs #6-#8 #9

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
29 changes: 19 additions & 10 deletions CONTEXT.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,21 +4,21 @@ Living document, unlike CLAUDE.md (timeless conventions). Read this at
the start of a session; update it at the end of any session that changes
project state (the `new-feature` and `pr-prep` skills enforce this).

_Last updated: 2026-07-13_
_Last updated: 2026-07-14_

## Active branches & worktrees

| Branch | Worktree | What it's for |
|---|---|---|
| `main` | `vectorDB/` (primary) | Stable; PRs merge here |
| `dev` | — | Integration branch |
| `banchmark` | — | Benchmark work (merged to main via PR #6) |
| `chore/ai-workflow` | `vectorDB/` (current) | This AI-workflow setup |

`banchmark` (PRs #6, #8) and `chore/ai-workflow` (PR #7) are merged and
can be deleted.

## In-flight work

- **AI workflow setup** (`chore/ai-workflow`): skills, agents, specs,
this file — pending PR.
None — working tree clean, all branches merged as of 2026-07-14.

## Recent decisions

Expand All @@ -30,6 +30,16 @@ _Last updated: 2026-07-13_
- **Spec-first workflow adopted** — features start from
[docs/specs/TEMPLATE.md](docs/specs/TEMPLATE.md) via the `new-feature`
skill; specs are the design record.
- **E2E benchmark baseline captured** (PR #6) — `cmd/bench` harness;
100k-vector baseline in
[bench/results/](bench/results/baseline_20260713-173909.json), tables
and version-tracking log in [bench/README.md](bench/README.md).
Headline: recall@10 = 0.681 at efSearch=50, flat recall curve points
at fogfish/hnsw graph-construction quality as the ceiling.
- **Maintenance landed via PR #8** — GitHub Actions CI (cgo + nocgo),
portable relative paths in the default config, and removal of the
legacy `db/`/`storage/` brute-force scaffold (CLAUDE.md updated in
the same merge).

## Next up

Expand All @@ -38,8 +48,7 @@ _Last updated: 2026-07-13_
needs real amd64 hardware for benchmarks.
- Consider insert-time vector normalization (cosine → dot product; halves
FLOPs, changes stored-data semantics) — needs a spec.
- Planned maintenance (branches not yet created): GitHub Actions CI
(cgo + nocgo, amd64 + arm64); remove the legacy `db/`/`storage/`
brute-force scaffold (update CLAUDE.md when it lands); portable
relative paths in the default config (`config.yaml` has absolute
macOS paths).
- Benchmark baseline flagged two index issues worth specs: `efSearch`
not clamped to `k` in `HNSWIndex.Search` (recall@100 = 0.40), and the
per-result Badger `Get` hydration tax in `Engine.Search` (7x QPS loss
at k=100).
Loading