You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 909cc1a
Browse filesBrowse the repository at this point in the historyBrowse files
Merge remote-tracking branch 'origin/main' into push-gateway-mesh-delivery
* origin/main:
feat(relay): enforce NIP-FI assertion+NIP-98 pairing on HTTP ingress (#7264)
fix(ci): run the admin disabled-mode DB test in the PostgreSQL lane (#7900)
🤖 docs: add pre-PR checklist and review guidance to AGENTS.md (#7897)
docs: specify durable data backfills (#7326)
docs(vision): add /buzz/v1 read endpoints to the protocol contract (#7879)
🤖 fix(justfile): point just staging at the current staging relay (#7881)
fix(relay-admin): make thread deletions atomic and fence expired action leases under row lock (#7853)
feat(desktop): relay admin console for the /api/admin/v1 operator surface (#4768)
fix(mobile): keep retired sections manager out of successor cache (#7873)
Signed-off-by: Tom Brow <[email protected]>
Copy file name to clipboardExpand all lines: AGENTS.md
+44-2Lines changed: 44 additions & 2 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -25,6 +25,48 @@ and runtime evidence answer different questions.
25
25
26
26
---
27
27
28
+
## Before opening a PR
29
+
30
+
Open work in progress as a draft PR. Mark it ready for review only when this
31
+
checklist holds for the current change. Scale it to what changed:
32
+
documentation-only changes need content, link, and diff checks plus human
33
+
confirmation, not app runs.
34
+
35
+
1.**Agent review ran** under [Reviewing](#reviewing), and its recommended
36
+
blockers were fixed or explicitly declined by the human author. Optional
37
+
suggestions do not gate readiness.
38
+
2.**An agent exercised the changed behavior.** Client changes: the affected flow
39
+
in the app, using the native app or a device when the behavior needs it
40
+
(browser or headless Playwright counts only for what it can exercise). Relay
41
+
changes: a local relay, exercising the changed events or endpoints. CLI or
42
+
tooling changes: the affected command or workflow. The human may skip this
43
+
step, for example for a small change or while iterating.
44
+
3.**A human then tested it themselves**: in the app, against the local relay
45
+
(through the app or `curl`), or by running the changed command. Agent testing
46
+
does not substitute. Agents give the human exact steps and what working looks
47
+
like, then wait for explicit confirmation. Never mark this step done yourself.
48
+
4.**Add `buzz-review-completed` to the PR description** once steps 1–3 hold. It
49
+
attests the checklist, and automated reviewers may skip review because of it.
50
+
If later edits change behavior, remove it and return the PR to draft until the
51
+
affected steps are redone.
52
+
53
+
## Reviewing
54
+
55
+
- Before reviewing, read [VISION.md](VISION.md), the `VISION_*.md` docs for the affected surface, and the PR's stated goal and linked
56
+
issue. Review the change against what it is trying to do.
57
+
- Check the change against the [Review-Proven Rules](#review-proven-rules): the defects reviewers here find most often.
58
+
- Judge minimalism, elegance, and correctness, aiming for 9/10 on each. A score
59
+
below 9 names the concrete defect and the fix.
60
+
- Recommend blocking only for concrete correctness, security, or agreed-contract
61
+
defects with a realistic failure scenario: state the defect, how it fails, and
62
+
the fix. Label everything else (nits, wording, speculative hardening,
63
+
out-of-scope improvements) as optional.
64
+
- Put all findings in the first review. Later reviews check prior blockers and
65
+
defects the fixes introduced; reopen other areas only on new evidence of a
66
+
material defect.
67
+
- Agents post reviews as comments, never Request Changes. Humans decide which
68
+
findings must be fixed.
69
+
28
70
## Ecosystem
29
71
30
72
Buzz spans five repos. This one (`block/buzz`) is the OSS source for the relay, desktop, mobile, and CLI. The others handle internal builds and deployment:
@@ -165,8 +207,8 @@ the same clusters and measured how often authors actually fix each class
165
207
once flagged: test-seam binding and unbounded-resource findings were fixed
166
208
**100%** of the time, swallowed-error findings **90%**, stale-state races
167
209
**70%** — these are not style opinions, they are defects authors agree
168
-
with on sight. Apply the rules**before writing code**; each cites the
169
-
PRs where reviewers litigated it.
210
+
with on sight. Authors apply them**before writing code**, and reviewers check
211
+
against them; each cites the PRs where reviewers litigated it.
170
212
171
213
1.**Every caught failure must leave a durable retry record or propagate.**
172
214
Never catch-log-and-return-success (opt-out revocation permanently
Copy file name to clipboardExpand all lines: VISION.md
+2Lines changed: 2 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -74,6 +74,8 @@ Buzz extends the standard Nostr event format with custom kind numbers for enterp
74
74
75
75
New message type? New kind integer. Zero breaking changes.
76
76
77
+
**The conversation is Nostr; Buzz adds accessories around it.** Messages, threads, reactions, and the rest of the conversation are signed Nostr events, and Nostr reads must cover all of that state, so a standards-leaning Buzz client stays possible. Around it, the relay serves accessories that don't fit that shape: Blossom media, git smart HTTP, workflow webhooks, derived unread counts, and namespaced `/buzz/v1/...` endpoints for backend-for-frontend reads (thread catch-up, sync) over the same authoritative store. We start with this rule and refine it as more cases show up.
0 commit comments