feat(replay-vision): rank the watch feed with jev behind a flag

[ksvat]

Problem

Layer 2 of the Jev watch-feed stack, on top of #107092. The base PR judges scanner windows hourly and caches the probabilities; nothing reads them yet.

Changes

• Teams whose vision-watch-feed-ranker flag is jev get a What to watch feed ranked on the cached Jev probabilities: highest probability first, viewed rows docked, newest as the tiebreak. Their cards say "The decision model judged this session worth watching."
• The two rankers are independent branches in the endpoint. weighted-score and jev-shadow teams run the existing rank_watch_feed_candidates exactly as today, and no Jev value enters that blend. Neither arm makes a model call at request time.
• A row without a cached probability (not yet swept, or the cache went cold) falls to the recency filler tier, so the feed never empties during rollout or a sweep outage.
• New jev_watchable reason kind: serializer choice with jev_probability, regenerated OpenAPI types, and the frontend copy case.

How did you test this code?

• New API test: catches a cached probability moving the feed for a weighted-score or jev-shadow team, and, under jev, broken probability ordering or a lost recency fallback for unjudged rows.
• Ranker unit tests live in the base PR (test_jev_watch_feed.py), including the viewed-row dock and the filler-tier reasons.
• Ran locally: the watch feed API test class, ruff, and repo-wide mypy. All passed.
• Not run: the full backend suite and frontend typecheck. CI covers them.

Release status

• No feature flag controls this change
• This change is behind a feature flag and is not available to users
• This change makes a previously flagged feature available to everyone

Automatic notifications

• Publish to changelog?

Docs update

None. The behavior is flag-gated and not user-visible yet.

🤖 Agent context

Autonomy: Human-driven (agent-assisted)

Agent: Claude Code (PostHog Desktop task), Claude Fable 5

• See the base PR for the design history. This layer holds only the read path and its contract changes, so the feed switch can be reviewed apart from the sweep machinery.
• Skills invoked: /writing-tests, /writing-user-facing-copy (reason copy), /improving-drf-endpoints, /writing-pr-descriptions.
• CodeRabbit local pass skipped (cloud task run).

---

Created with PostHog Desktop

🤖 Generated with Claude Code

[github-actions]

🤖 CI report

⚠️ Trunk lane — backend Python lane

This PR is assigned to the backend Python lane. It runs backend Python tests and may merge in parallel with PRs in other lanes.

⚠️ Complexity (TypeScript) — 3 functions above the limit (max 33)

Cyclomatic complexity above the limit in changed typescript files (10 for production files, 15 for test files). Warn only: worth simplifying when you next touch these functions.

Function	Location	Complexity	Limit
watchReasonCopy	products/replay_vision/frontend/replay_scanners/components/WatchFeedCard.tsx:43	33	10
WatchFeedCard	products/replay_vision/frontend/replay_scanners/components/WatchFeedCard.tsx:206	17	10
watchCardHeadline	products/replay_vision/frontend/replay_scanners/components/WatchFeedCard.tsx:163	15	10

✅ Duplication (Python) — clean

New Python code duplication introduced by this branch. Fails at 70+ tokens in app code, or 150+ tokens when both copies live in test files. Advisory while the gate proves itself: extract a shared helper instead of copying.

✅ Duplication (TypeScript) — clean

New TypeScript code duplication introduced by this branch. Fails at 70+ tokens in app code, or 150+ tokens when both copies live in test files. Advisory while the gate proves itself: extract a shared helper instead of copying.

🚨 Comment density — 15% of added code lines are comments (18 of 118)

This section warns when comments are more than 3% of the code lines a PR adds, and alerts above 6%. Before agent-assisted PRs, the typical share was about 2%. Only full-line comments count. Docstrings, generated files, snapshots, migrations, and workflow files are left out.

Comments that restate the code, record how the change came about, or narrate the next line add noise for the next reader. Keep the comments that explain a reason the code cannot show, and remove the rest. See .agents/skills/writing-code-comments/SKILL.md for the house rules.

Files with the most added comment lines:

File	Comment lines	Added lines
products/replay_vision/backend/tests/test_api.py	11	70
products/replay_vision/backend/api/scanners.py	7	46

This check does not block merging. It updates on every push and clears when the share drops.

⚠️ Bundle size — 🔺 +83 B (+0.0%)

Uncompressed size of every built .js bundle, compared against the base branch.

Total: 68.87 MiB · 🔺 +83 B (+0.0%)

No file changed by more than 1000 B.

_{Posted automatically by build-bundle-size-report · uncompressed bytes from dist-report}

✅ Eager graph — within budget

How much code each root ships on the eager path — downloaded and parsed before the surface is interactive. Measured from the esbuild output chunks (post-tree-shake, static imports only); lazy import() / React.lazy chunks are not counted.

Root	Eager (shipped)	Δ vs base	Budget
entry (logged-out pages, app bootstrap) src/index.tsx	1.57 MiB · 22 files	no change	█████████░ 85.5% of 1.84 MiB
logged-out boot: index + App + bootApp (preloaded by every page, including /login) src/index.tsx + src/scenes/App.tsx + src/scenes/bootApp.ts	3.51 MiB · 629 files	no change	█████████░ 87.2% of 4.03 MiB
authenticated shell (every logged-in page) src/scenes/AuthenticatedShell.tsx	7.33 MiB · 2,332 files	no change	█████████░ 87.9% of 8.34 MiB

🟢 node_modules/monaco-editor/ stays out of src/index.tsx
🟢 src/lib/components/ActivityLog/describers stays out of src/index.tsx
🟢 [object Object] stays out of src/index.tsx
🟢 [object Object] stays out of src/index.tsx
🟢 node_modules/monaco-editor/ stays out of src/index.tsx + src/scenes/App.tsx + src/scenes/bootApp.ts
🟢 src/layout/navigation-3000/navigationLogic.tsx stays out of src/index.tsx + src/scenes/App.tsx + src/scenes/bootApp.ts
🟢 src/scenes/dashboard/dashboardLogic.tsx stays out of src/index.tsx + src/scenes/App.tsx + src/scenes/bootApp.ts
🟢 src/lib/lemon-ui/LemonMarkdown/ stays out of src/index.tsx + src/scenes/App.tsx + src/scenes/bootApp.ts
🟢 src/lib/components/RichContentEditor/ stays out of src/index.tsx + src/scenes/App.tsx + src/scenes/bootApp.ts
🟢 src/lib/components/CodeSnippet/ stays out of src/index.tsx + src/scenes/App.tsx + src/scenes/bootApp.ts
🟢 src/taxonomy/core-filter-definitions-by-group.json stays out of src/index.tsx + src/scenes/App.tsx + src/scenes/bootApp.ts
🟢 node_modules/monaco-editor/ stays out of src/scenes/AuthenticatedShell.tsx
🟢 src/lib/components/ActivityLog/describers stays out of src/scenes/AuthenticatedShell.tsx
🟢 [object Object] stays out of src/scenes/AuthenticatedShell.tsx
🟢 src/scenes/session-recordings/player/sessionRecordingPlayerLogic.ts stays out of src/scenes/AuthenticatedShell.tsx
🟢 [object Object] stays out of src/scenes/AuthenticatedShell.tsx
🟢 [object Object] stays out of src/scenes/AuthenticatedShell.tsx
🟢 [object Object] stays out of src/scenes/AuthenticatedShell.tsx

Largest files eagerly shipped from src/index.tsx

Size	File
126.8 KiB	../node_modules/.pnpm/react-dom@18.3.1_react@18.3.1/node_modules/react-dom/cjs/react-dom.production.min.js
24.6 KiB	../node_modules/.pnpm/buffer@6.0.3/node_modules/buffer/index.js
6.3 KiB	../node_modules/.pnpm/react@18.3.1/node_modules/react/cjs/react.production.min.js
4.5 KiB	../node_modules/.pnpm/@jspm+core@2.1.0/node_modules/@jspm/core/nodelibs/browser/process.js
3.9 KiB	../node_modules/.pnpm/scheduler@0.23.2/node_modules/scheduler/cjs/scheduler.production.min.js
1.4 KiB	../node_modules/.pnpm/base64-js@1.5.1/node_modules/base64-js/index.js
1.3 KiB	src/index.tsx
1.3 KiB	src/RootErrorBoundary.tsx
912 B	../node_modules/.pnpm/ieee754@1.2.1/node_modules/ieee754/index.js
854 B	src/scenes/ChunkLoadErrorBoundary.tsx

Largest files eagerly shipped from src/index.tsx + src/scenes/App.tsx + src/scenes/bootApp.ts

Size	File
301.8 KiB	../node_modules/.pnpm/posthog-js@1.434.14_@types+react@18.3.27_react@18.3.1/node_modules/posthog-js/dist/module.mjs
216.0 KiB	../node_modules/.pnpm/@posthog+icons@0.38.0_react-dom@18.3.1_react@18.3.1__react@18.3.1/node_modules/@posthog/icons/dist/posthog-icons.es.js
126.8 KiB	../node_modules/.pnpm/react-dom@18.3.1_react@18.3.1/node_modules/react-dom/cjs/react-dom.production.min.js
100.4 KiB	src/lib/api.ts
88.4 KiB	src/products.tsx
69.4 KiB	src/lib/lemon-ui/icons/icons.tsx
40.1 KiB	src/lib/utils/eventUsageLogic.ts
38.7 KiB	../node_modules/.pnpm/@dnd-kit+core@6.0.8_react-dom@18.3.1_react@18.3.1__react@18.3.1/node_modules/@dnd-kit/core/dist/core.esm.js
33.9 KiB	../node_modules/.pnpm/kea@4.0.0-pre.6_patch_hash=139b8d1f1304f9d9da452a9a1244c94ea679dbcb85687d8999563146879fb6f5_react@18.3.1/node_modules/kea/lib/index.cjs.js
28.4 KiB	src/scenes/scenes.ts

Largest files eagerly shipped from src/scenes/AuthenticatedShell.tsx

Size	File
301.8 KiB	../node_modules/.pnpm/posthog-js@1.434.14_@types+react@18.3.27_react@18.3.1/node_modules/posthog-js/dist/module.mjs
271.7 KiB	src/taxonomy/core-filter-definitions-by-group.json
216.0 KiB	../node_modules/.pnpm/@posthog+icons@0.38.0_react-dom@18.3.1_react@18.3.1__react@18.3.1/node_modules/@posthog/icons/dist/posthog-icons.es.js
153.7 KiB	../node_modules/.pnpm/re2js@0.4.1/node_modules/re2js/build/index.esm.js
126.8 KiB	../node_modules/.pnpm/react-dom@18.3.1_react@18.3.1/node_modules/react-dom/cjs/react-dom.production.min.js
100.4 KiB	src/lib/api.ts
98.5 KiB	../packages/quill/packages/quill/dist/index.js
93.3 KiB	../node_modules/.pnpm/prosemirror-view@1.40.1/node_modules/prosemirror-view/dist/index.js
90.6 KiB	../node_modules/.pnpm/@tiptap+core@3.20.6_@tiptap+pm@3.20.6/node_modules/@tiptap/core/dist/index.js
88.4 KiB	src/products.tsx

_{Posted automatically by check-eager-graph · sizes are eager output bytes (shipped, post-tree-shake) from the esbuild metafile · part of #32479}

✅ Toolbar bundle — eager 2.16 MiB within budget

What the toolbar ships to customer pages, measured from the esbuild output (minified, post-tree-shake). The eager set is the entry plus everything statically imported from it — fetched before any feature runs; deferred chunks load lazily. The eager guardrail is 5.72 MiB. Each output file must also stay below 10 MB, where CloudFront stops compressing it. The module boundary is enforced separately by check-toolbar-graph.

Metric	Size	Δ vs base	Budget
Eager (shipped) entry + static imports	2.16 MiB · 19 files	no change	████░░░░░░ 37.7% of 5.72 MiB
Deferred (lazy)	2.10 MiB · 44 files	no change	n/a — loads on demand
Loader dist/toolbar.js	1.2 KiB	no change	█░░░░░░░░░ 6.0% of 19.5 KiB

Largest eagerly-shipped chunks

Size	File
800.5 KiB	dist/toolbar/toolbar-app-JG6G2U7S.css
651.5 KiB	dist/toolbar/chunk-chunk-DJ3DVXIY.js
259.4 KiB	dist/toolbar/chunk-chunk-CV2VU6SQ.js
138.3 KiB	dist/toolbar/chunk-chunk-TUPNHXOT.js
131.8 KiB	dist/toolbar/chunk-chunk-FDH2IBXT.js
75.2 KiB	dist/toolbar/toolbar-app-QSMPSM2L.js
69.0 KiB	dist/toolbar/chunk-chunk-TSAL54PB.js
35.6 KiB	dist/toolbar/chunk-chunk-P5FAEBEQ.js
21.0 KiB	dist/toolbar/chunk-chunk-W5LHFGNZ.js
6.8 KiB	dist/toolbar/chunk-chunk-DV7IWQNF.js

_{Posted automatically by check-toolbar-size · sizes are toolbar output bytes (shipped, post-tree-shake) from the esbuild metafile}

✅ Dist folder size — 🔺 +1.8 KiB (+0.0%)

Total size of the built frontend/dist folder (all assets), compared against the base branch.

Total: 946.07 MiB · 🔺 +1.8 KiB (+0.0%)

ℹ️ MCP UI apps size — 33 app(s), 17631.0 KB JS

Built size of each MCP UI app (main.js + styles.css).

App	JS	CSS
debug	597.9 KB	196.2 KB
action	454.1 KB	196.2 KB
action-list	564.2 KB	196.2 KB
cohort	453.1 KB	196.2 KB
cohort-list	563.2 KB	196.2 KB
email-template	452.9 KB	196.2 KB
error-details	469.6 KB	196.2 KB
error-issue	453.8 KB	196.2 KB
error-issue-list	564.1 KB	196.2 KB
experiment	561.3 KB	196.2 KB
experiment-list	564.9 KB	196.2 KB
experiment-results	566.3 KB	196.2 KB
feature-flag	566.8 KB	196.2 KB
feature-flag-list	570.5 KB	196.2 KB
feature-flag-testing	457.3 KB	196.2 KB
inline-scan	453.6 KB	196.2 KB
insight-actors	562.3 KB	196.2 KB
invite-email-preview	452.3 KB	196.2 KB
llm-costs	559.3 KB	196.2 KB
session-recording	455.3 KB	196.2 KB
survey	454.7 KB	196.2 KB
survey-global-stats	561.9 KB	196.2 KB
survey-list	564.9 KB	196.2 KB
survey-stats	561.9 KB	196.2 KB
trace-span	453.5 KB	196.2 KB
trace-span-list	564.1 KB	196.2 KB
vision-observation-list	563.3 KB	196.2 KB
workflow	453.4 KB	196.2 KB
workflow-list	563.5 KB	196.2 KB
loops-review	457.8 KB	196.2 KB
query-results	774.1 KB	196.2 KB
render-ui	857.4 KB	196.2 KB
visual-review-snapshots	457.9 KB	196.2 KB

✅ Playwright — all passed

All tests passed.

View test results →

⚠️ Backend coverage — 96.0% of changed backend lines covered — 25 uncovered

🧪 Backend test coverage

Patch coverage — changed backend lines (products + core): ███████████████████░ 96.0% (667 / 692)

File	Patch	Uncovered changed lines
products/ml_inference/backend/facade/api.py	50.0%	16
products/replay_vision/backend/temporal/jev_watch_rank/schedule.py	87.5%	19
products/replay_vision/backend/temporal/jev_watch_rank/workflow.py	87.5%	26, 33
products/replay_vision/backend/temporal/jev_watch_rank/activities.py	89.3%	115–116, 141–142, 155–156, 158–159, 163, 221, 223–224, 250
products/replay_vision/backend/jev_watch_feed.py	96.1%	375–377, 399–401, 403, 425

🤖 Agents: add a test covering the lines above, or note why under "How did you test this code?". Machine-readable gap list: the patch-coverage artifact on this run (gh run download 36495114150 -n patch-coverage), or the coverage-data block at the end of this comment.

Per-product line coverage (touched products)

Product	Coverage	Lines
platform_features	██░░░░░░░░░░░░░░░░░░ 12.1%	7 / 58
warehouse_sources_queue	██████░░░░░░░░░░░░░░ 29.1%	92 / 316
demo	███████████░░░░░░░░░ 52.8%	1,411 / 2,673
data_tools	████████████░░░░░░░░ 61.2%	90 / 147
aeo	██████████████░░░░░░ 70.8%	472 / 667
ai_gateway	███████████████░░░░░ 75.0%	9 / 12
batch_exports	████████████████░░░░ 81.2%	21,475 / 26,449
apm	█████████████████░░░ 84.1%	1,306 / 1,553
cdp	██████████████████░░ 88.2%	4,545 / 5,155
ml_inference	██████████████████░░ 88.2%	510 / 578
mcp_analytics	██████████████████░░ 88.9%	4,927 / 5,540
product_tours	██████████████████░░ 89.3%	1,331 / 1,491
dashboards	██████████████████░░ 89.5%	6,855 / 7,657
data_warehouse	██████████████████░░ 90.0%	14,019 / 15,581
signals	██████████████████░░ 90.0%	56,092 / 62,306
notebooks	██████████████████░░ 90.2%	15,287 / 16,945
cohorts	██████████████████░░ 90.4%	8,420 / 9,316
streamlit_apps	██████████████████░░ 90.6%	2,623 / 2,895
tasks	██████████████████░░ 91.0%	74,108 / 81,470
managed_warehouse	██████████████████░░ 91.0%	10,252 / 11,263
data_modeling	██████████████████░░ 91.4%	10,504 / 11,498
exports	██████████████████░░ 91.7%	9,676 / 10,556
engineering_analytics	██████████████████░░ 91.7%	11,032 / 12,030
ai_training	██████████████████░░ 92.2%	356 / 386
business_knowledge	██████████████████░░ 92.2%	7,684 / 8,330
conversations	███████████████████░ 92.5%	28,734 / 31,062
early_access_features	███████████████████░ 92.6%	1,341 / 1,448
managed_migrations	███████████████████░ 92.7%	1,581 / 1,705
visual_review	███████████████████░ 92.8%	9,247 / 9,969
canvas	███████████████████░ 92.8%	6,877 / 7,409
approvals	███████████████████░ 93.0%	3,974 / 4,271
mcp_registry	███████████████████░ 93.1%	1,670 / 1,794
error_tracking	███████████████████░ 93.2%	15,928 / 17,097
notifications	███████████████████░ 93.2%	1,145 / 1,229
slack_app	███████████████████░ 93.2%	13,677 / 14,674
stamphog	███████████████████░ 93.2%	7,885 / 8,456
surveys	███████████████████░ 93.3%	6,571 / 7,040
context_layer	███████████████████░ 93.9%	3,415 / 3,638
web_analytics	███████████████████░ 93.9%	21,687 / 23,096
alerts	███████████████████░ 94.0%	8,569 / 9,114
billing_alerts	███████████████████░ 94.1%	2,094 / 2,226
mcp_store	███████████████████░ 94.4%	8,940 / 9,472
ai_observability	███████████████████░ 94.6%	24,063 / 25,440
wizard	███████████████████░ 94.7%	6,150 / 6,496
reminders	███████████████████░ 94.8%	760 / 802
workflows	███████████████████░ 94.8%	14,332 / 15,113
review_hog	███████████████████░ 95.0%	11,507 / 12,119
endpoints	███████████████████░ 95.1%	9,206 / 9,681
annotations	███████████████████░ 95.1%	817 / 859
customer_analytics	███████████████████░ 95.2%	25,078 / 26,349
legal_documents	███████████████████░ 95.2%	2,311 / 2,427
marketing_analytics	███████████████████░ 95.3%	19,322 / 20,274
posthog_ai	███████████████████░ 95.3%	2,488 / 2,610
experiments	███████████████████░ 95.4%	32,458 / 34,023
actions	███████████████████░ 95.5%	756 / 792
logs	███████████████████░ 95.5%	15,399 / 16,130
data_catalog	███████████████████░ 95.5%	4,401 / 4,606
tracing	███████████████████░ 95.6%	3,536 / 3,699
autoresearch	███████████████████░ 95.7%	8,481 / 8,865
growth	███████████████████░ 95.7%	11,228 / 11,734
messaging	███████████████████░ 95.8%	3,798 / 3,963
skills	███████████████████░ 95.8%	6,972 / 7,274
replay_vision	███████████████████░ 95.9%	27,818 / 28,999
product_analytics	███████████████████░ 96.0%	28,470 / 29,647
access_control	███████████████████░ 96.3%	7,112 / 7,386
revenue_analytics	███████████████████░ 96.4%	1,876 / 1,946
user_interviews	███████████████████░ 96.5%	2,859 / 2,963
feature_flags	███████████████████░ 96.5%	25,556 / 26,473
warehouse_sources	███████████████████░ 97.2%	456,705 / 469,652
data_quality	████████████████████ 97.6%	7,587 / 7,774
security	████████████████████ 97.9%	1,202 / 1,228
links	████████████████████ 97.9%	234 / 239
metrics	████████████████████ 98.0%	4,084 / 4,166
analytics_platform	████████████████████ 98.3%	2,784 / 2,833
pulse	████████████████████ 98.5%	2,043 / 2,075
live_debugger	████████████████████ 99.2%	626 / 631
field_notes	████████████████████ 99.4%	172 / 173

Report-only. Patch coverage = changed backend lines covered vs origin/master. Sorted lowest first.
Known gaps: lines covered only by Temporal tests show as uncovered; core line numbers may drift if master changed the same file.

[coderabbitai]

Navigate logical layers of code changes, visualize relationships, and explore their blast radius.

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Use the following commands to manage reviews:

• @coderabbitai resume to resume automatic reviews.
• @coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

• ▶️ Resume reviews
• 🔍 Trigger review

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository: PostHog/posthog/.coderabbit.yaml

Review profile: QUIET

Plan: Enterprise

Run ID: b7cad819-ed8e-49f7-8296-0eb8b16c6bec

📥 Commits

Reviewing files that changed from the base of the PR and between e73fb1d and 721ca3c.

⛔ Files ignored due to path filters (1)

• products/replay_vision/frontend/generated/api.schemas.ts is excluded by !**/generated/**

📒 Files selected for processing (1)

• services/mcp/src/api/generated.ts

Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 9 remain after this review.

---

📝 Walkthrough

Walkthrough

The watch-feed API selects JEV ranking when the team’s ranker is jev. Other ranker modes use the existing candidate ranker. Watch-feed reasons include jev_watchable and an optional nullable probability. The frontend adds explanatory copy for the new reason. A test covers behavior across weighted-score, jev-shadow, and jev.

Priority: ➖ Normal

Merge Risk: 🔵 Low · up to 721ca

Teams assigned to Jev can see more than the documented three recency-filler rows while judgments are unavailable. The default remains weighted-score, so this is bounded to the Jev rollout; address or accept the cap difference before broadening it.

Security Architecture Review

Security architecture risk: 🔵 Low · up to 721ca

Access checks still apply before ranking, and no cross-tenant exposure was found. When scores are unavailable, however, the feed can return more routine cards than its documented fallback permits.

Retained concerns

• Low · reliability · observed: On the JEV arm, missing cached scores can cause routine filler rows to occupy the full requested limit, rather than the existing three-item fallback. This changes the feed’s documented behavior during cache failure or rollout.

Security review details

Security Blast Radius

• inferred — The changed ordering and additional reason metadata can affect authorized watch-feed readers on JEV-enabled teams; the examined path does not expand the candidate set beyond the existing team and access filters.

Trust Boundaries and Controls

• observed — Caller-provided feed filters are validated, scanner IDs are intersected with readable IDs, and observation access is checked before cached ranks influence the response.

Resilience and Maintainability Implications

• observed — A cache read failure preserves endpoint availability through recency fallback, but the JEV fallback does not preserve the weighted arm’s three-item filler bound.

🚥 Pre-merge checks | ✅ 1

✅ Passed checks (1 passed)

Check name	Status	Explanation
Description check	✅ Passed	The description is complete and aligned with the template. It explains the problem, user-visible changes, testing performed and not performed, feature-flag release status, documentation status, and ag…

✨ Finishing Touches

📝 Generate docstrings

• Commit to this branch
• Create a new PR

🧪 Generate unit tests (beta)

• Commit to this branch
• Create a new PR

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository: PostHog/posthog/.coderabbit.yaml

Review profile: QUIET

Plan: Enterprise

Run ID: 2a11e840-749e-487e-a3f6-43d0d33a1f96

📥 Commits

Reviewing files that changed from the base of the PR and between 7df6c53 and 7f97ce8.

⛔ Files ignored due to path filters (1)

• products/replay_vision/frontend/generated/api.schemas.ts is excluded by !**/generated/**

📒 Files selected for processing (4)

• products/replay_vision/backend/api/scanners.py
• products/replay_vision/backend/tests/test_api.py
• products/replay_vision/frontend/replay_scanners/components/WatchFeedCard.tsx
• services/mcp/src/api/generated.ts

Included review availability: Your plan provides up to 12 included reviews per hour; 10 remain after this review.

[posthog]

👋 Visual changes detected for this PR.

Review and approve in PostHog Visual Review

If these changes are unexpected, they may be caused by a flaky test or a broken snapshot on master. Don't approve — rerun the job or wait for a fix.

[trunk-io]

   

View Full Report ↗︎ ⋅ Docs

[github-actions]

🦔 Hogbox preview · ✅ ready

▶ Open the preview

🔑 Login	test@posthog.com / 12345678 (demo data)
🧩 Running	this PR's backend and frontend, on the PostHog :master base
🔗 Link	stable across rebuilds — a re-push swaps the box underneath, the URL stays
🔒 Access	tailnet only (PostHog VPN)
🛠️ Admin	inspect & debug state in hogland
💤 Idle	sleeps after ~30 min idle (snapshot to S3, zero node cost) and wakes on your next visit in ~30s, behind a brief "waking up" screen

_{commit b720826 · box box-565f92c03600 · ready in 771s (push → usable) · build log · rebuilds on every push, torn down on close}

[stamphog]

Not approved — this change needs a human reviewer.

Re-add the stamphog label to request another review once you have addressed this.

CodeRabbit's Major finding on scanners.py — that a cold Jev cache lets the new ranking branch return up to the full page-size limit of recency-filler rows instead of the feed's normal 3-row floor for a window with no findings — is marked outdated but is still present in the current code: rank_watch_feed_by_jev has no equivalent of the weighted ranker's filler trim, and the call site slices straight by params["limit"] with no floor applied. That's a genuine, unaddressed behavioral regression the PR itself says will happen during rollout.

• Author wrote 75% of the modified lines and has 37 merged PRs in these paths (familiarity STRONG).
• Unaddressed CodeRabbit comment on products/replay_vision/backend/api/scanners.py: the jev ranking branch can return up to params["limit"] recency-filler rows when the cache is cold, instead of trimming to the feed's normal no-findings floor of 3 rows (see watch_feed.py's _trim_filler / WATCH_FEED_MIN_ITEMS, which rank_watch_feed_by_jev has no equivalent of).

Gate mechanics and policy version

Gate		Result
prerequisites	✓	all clear
deny-list	✓	no deny categories matched
size	✓	52L, 2F substantive, 150L/5F incl. docs/generated/snapshots — within ceiling
tier	✓	T1-agent / T1c-medium (150L, 5F, two-areas, feat)
stamphog 2.2.0		.stamphog/policy.yml @ 1bd808e · reviewed head 1bd808e

[TueHaulund]

🤖 Agent-drafted, reviewed by Tue.

CodeRabbit is right here. The jev path never drops filler rows like the weighted path does. If the cache is empty, a team gets 20 "new since you last looked" cards instead of 3. That also makes the two variants hard to compare. Wrapping this in _trim_filler(...) before the slice should fix it.

[ksvat]

Fixed in aca1032 with exactly that shape — the jev result now runs through _trim_filler, so both arms keep the same 3-row floor for windows without findings and stay comparable in the experiment. Unit test plus a call-site assertion cover it.

[TueHaulund]

🤖 Agent-drafted, reviewed by Tue.

A jev_watchable card always shows the scan's notability_reason, even when that reason is "nothing stands out". The weighted path only shows it when the scan found something notable. Could the jev path do the same?

[ksvat]

Fixed in aca1032: the jev card now carries notability_reason only when the scan's own notability clears NOTABLE_MIN_SCORE, the same bar the weighted path applies, so a card Jev rated watchable relative to a dull window can no longer lead with "nothing stands out". Covered by a unit test and the end-to-end reason assertion in the API tests.

[arnohillen]

list_watch uses the jev ranker in all regions, but the sweep writes ranks only where decision_api.decisions_available_here() is true (activities.py:110), so an EU team with the flag on gets only recency filler rows. Make watch_feed_ranker return "weighted-score" when decisions_available_here() is false, and add a test for it.

[ksvat]

Deliberately deferring this while the experiment is internal-only: enrollment is explicitly team-targeted (PostHog's own team, US cloud), so no EU team can land on the jev arm, and with the filler floor fixed a misconfigured team would now see 3 recency rows rather than a full page. Agreed the region gate belongs in watch_feed_ranker before the flag opens beyond internal enrollment — it's recorded in the same TODO as the team-2 pin (constants.py, base PR), so both internal-only assumptions get unwound together.

A new stamphog review started for this PR — the fresh verdict replaces this approval.

A new stamphog review started for this PR — the fresh verdict replaces this approval.

[stamphog]

Approved.

This is a flag-gated read path in a contained product area. The earlier reviewer concerns (filler trimming, notability reason) are fixed in the tree. The unresolved EU-region point degrades safely to recency rows and is deferred while enrollment is internal-only. The author has STRONG familiarity with the code, and the change is easy to revert.

• Author wrote 75% of the modified lines and has 37 merged PRs in these paths (familiarity STRONG).
• Non-blocking: @arnohillen's unresolved thread notes the jev ranker isn't region-gated (EU teams would only get recency filler). Author has deferred it to before the flag widens beyond internal teams; make sure that follow-up lands.

Gate mechanics and policy version

Gate		Result
prerequisites	✓	all clear
deny-list	✓	no deny categories matched
size	✓	52L, 2F substantive, 152L/5F incl. docs/generated/snapshots — within ceiling
tier	✓	T1-agent / T1c-medium (152L, 5F, two-areas, feat)
stamphog 2.3.0		.stamphog/policy.yml @ b720826 · reviewed head b720826