Split AGENTS.md reference sections into docs/; keep thin agent constitution

Extracts Playwright harness tables and Dark Factory testing standard to
docs/testing/, and dated incident writeups to docs/lessons/, bringing
AGENTS.md under the 32KiB context budget agent harnesses eagerly load.

Co-Authored-By: Claude Fable 5 <[email protected]>
This commit is contained in:
thePR0M3TH3ANandClaude Fable 5 committed 2026-08-12 15:20:08 -04:00
1 parent 6eb89afcec
commit 911afb1c05
7 files changed
+283 -247

No files matched your search

@@ -0,0 +1,10 @@
# Cloudflare/S3 Upload-Path Gotchas (audited 2026-06-16)
> Extracted verbatim from `AGENTS.md` §4a, which retains a short summary. Dated incident writeup from the 2026-06-16 upload-path audit.
* **Two near-identical upload services exist** — `js/services/r2Service.js` (Cloudflare R2) and `js/services/s3UploadService.js` (generic S3). They share `buildR2Key`, `resolveUploadIdentifier`, and the magnet build but are *copies*. **Any fix to one must be mirrored in the other** or they silently drift.
* **Storage-key collisions = data loss — FIXED 2026-06-16.** `buildR2Key` (`js/r2.js`) builds `u/<npub>/<namespace>/<slug>.<ext>` where `namespace` used to fall back to the literal `"uploads"` and `slug` to `"video"` (e.g. a non-ASCII filename, or no info-hash) — so a URL-first upload with no info-hash and a duplicate filename overwrote the previous object while the old note still pointed at that URL. WebTorrent uploads were always safe (info-hash namespaces the key = content-addressed). Fix: `computeStorageContentHash(file)` (`js/r2.js`) derives a content-based namespace when no info-hash is available (full SHA-256 for ≤512 MB, sampled metadata+edge fingerprint above that to avoid buffering huge files), wired into **both** `r2Service` and `s3UploadService`. Test: `tests/storage-key-collision.test.mjs`. **If you add a third upload path, reuse `computeStorageContentHash`.**
* **Whole-file hash before upload.** When the modal doesn't pre-seed an info-hash, `resolveUploadIdentifier` runs `calculateTorrentInfoHash(file)`, which reads the *entire* file before the upload begins and before any progress is emitted — on multi-GB files this looks like a hang. Emit a "preparing/hashing" status or parallelize.
* **Publish can report success unconfirmed.** `publishEventToRelay` (`js/nostrPublish.js`) has an optimistic-success fallthrough when the relay handle exposes no `on()`/`then()` — a note may show "Published" with no relay ACK. Don't tighten blindly (legacy `seen`-only relays rely on it), but log it.
* **Silent partial failures.** Thumbnail and `.torrent` upload failures are caught and `warn`-only; the video note still publishes with no user signal. Empty (0-byte) files hit `CompleteMultipartUpload` with `Parts:[]` and surface a cryptic error — guard `file.size === 0`.
* **Object cleanup (deletes/edits).** `R2Service.deleteVideoStorage()` removes the backing objects so "deleted" videos aren't left publicly downloadable. Contract to preserve: it's **best-effort and never throws** (must not block note deletion), it derives keys via `collectVideoStorageKeys()` and **only ever touches URLs under the owner's bucket base** (never external/3rd-party URLs), and it skips cleanly when storage is locked/credential-less. Wired into delete (`app.handleDeleteModalConfirm`) and edit-with-replaced-URL (`editModalController.handleSubmit`). Edit cleanup intentionally does **not** delete thumbnails (the new note may still reference them) and only fires when both old and new URLs are present and different.
+15
View File
@@ -0,0 +1,15 @@
# NIP-07 Signer Reliability & Encrypted-List Loading (hard-won, 2026-06-16)
> Extracted verbatim from `AGENTS.md` §17, which retains a short summary. Dated incident writeup; see also `docs/KNOWN_BUGS.md` #0.
The single biggest source of "DMs / hashtags / watch-history / block & subscription lists won't load after login" is **not** bitvid — it's an **unresponsive NIP-07 signer**. The extension's MV3 background service-worker can die or its content-script↔worker channel can orphan, after which raw `window.nostr` calls hang forever. No client change can force a dead signer to answer.
* **Diagnose first, don't guess.** Run a single raw `window.nostr` probe in the page console (`getPublicKey → nip04.encrypt → nip04.decrypt`) that bypasses bitvid entirely. If *that* hangs, it's the extension/environment. Recommend a well-maintained signer (nos2x, Alby); KeysBand's dead worker was the root cause and switching to nos2x fixed everything.
* **Resilience invariants — do not regress these** (see `docs/KNOWN_BUGS.md` #0 for the full history and the files):
1. **Cap relay fan-out** (`js/nostr/toolkit.js` `capReadRelays`, ≤8, user-relays-first + 2 reserved default slots). An uncapped cold-login REQ storm to ~20 dead NIP-65 relays starves the single-threaded signer's postMessage round-trips.
2. **Circuit breaker** on the NIP-07 channel (`js/nostr/nip07Permissions.js`): after N consecutive *timeouts* fast-fail instead of hanging ~15s each. Channel-death errors must count toward opening (not reset it); interactive permission prompts bypass it; one periodic probe detects recovery.
3. **Never swallow a transient decrypt error as an empty result.** Re-throw channel-death/timeout sub-errors so the retry path runs — returning `[]` turns "signer is slow" into "user has no blocks/lists" and kills retries.
4. **Generous decrypt budget** (~25–30s/call, ~60s backoff cap); a 6s timeout kills slow-but-responsive signers mid-decrypt. Handshake variant timeout is ~20s for the same reason.
5. **One actionable user notice** (`js/utils/signerHealthNotice.js`) after a few timeouts with `signerStatus: "present"` — not a silent forever-retry.
* **Don't render into closed modals at login.** Once a responsive signer makes decryption instant, eagerly populating every profile panel at login (friends avatars, subscriptions, blocks, DM summaries) freezes the main thread ~10–15s. Populate each pane lazily on open (`selectPane`) and gate data-change re-render listeners on "is the view open".
* **Deterministic testing.** Reproduce signer-dependent bugs headlessly with a fake `window.nostr` (Playwright `exposeBinding` + `addInitScript`) + a mock relay + configurable channel models (healthy/slow/overload/dead, latency override). See `scripts/perf/nip07-channel-sim.mjs`.