Repository navigation
Conversation
🧙 Wizard CIRun the Wizard CI and test your changes against wizard-workbench example apps by replying with a GitHub comment using one of the following commands: Test all apps:
Test all apps in a directory:
Test an individual app:
Show more apps
Test against a wizard branch:
Add Results will be posted here when complete. |
The rewrite files pointed at a coarse-name list that only the Resend mapping has. Step 7 now lists the ten names, and each rewrite file points there. Each rewrite file now says whether reply-to takes one address, a list, or both, as the SDK heads do. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A client with the secret key loads flag definitions from its host with it: Go and Ruby when they start, Python and .NET when the app evaluates a flag. Reusing the app's client behind a reverse proxy handed the key to the proxy. Step 7 now gives email its own client on PostHog's host behind a proxy, for every SDK, in place of the Go and Ruby special cases. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The step 7 proxy test matched only the separate-client wording, so dropping "leave the secret key off the app's existing client" still passed. Assert that clause too; it fails without it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A send of a template stored at the provider now moves by bringing the email into the code: its HTML and text become a file next to the send, the code renders the variables with the provider's escaping, and the site sends the code form. A template made in PostHog, sent by its ID, is used only when the person asks for one. - template-to-code.md: the check, where the email goes, how each language renders it without a new dependency, and the escaping rule. - Steps 3, 7 and 9: the new default, the template form on request, and the report items for emails brought into the code. - SendGrid: Handlebars in the code, including the helpers the template form couldn't take. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…pty Text as left out Step 3 reads only the Check part of the template file, so the rule that keeps a template whose provider sent no text part moves there. Resend's Go SDK leaves an empty Text out of the request, so Resend made that text part from the HTML, and such a site moves. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ain text off A dashboard copy shows no plain text both for a template without one and for one whose plain text is an empty string. Ask which, or read the template from the API; when nobody can tell, the site stays. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…tion and keep the answer A dashboard copy has no key to read the template from the API, so the question offers to make one. The answer goes into the template's plan note, so a copy made again in Step 7 asks nothing new. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…nly in Step 3 The template's note records its plain text however the template was read, so Step 7 never asks again, and the key is offered only in Step 3, before Step 5 asks the person to delete it. A site stays whenever neither the answer nor the API read settles it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
PostHog's bulk add takes only bounce, complaint or manual, and writes its own reason text, such as "Imported (bounced)". The removal guard now keys on that prefix, and the import maps each Resend origin to the reason value. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…them PostHog's list records no source an app could check before a removal, a bulk add skips an address removed earlier, and an app can't call the import tool. PostHog blocks a transactional send only to an address that hard-bounced, and records that itself. So each such call becomes a "Before cutover" item. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
PostHog blocks a transactional send only to an address that hard-bounced, so the weekly re-import limits mail to bounced addresses, not complaints. A call that removes an address becomes an "After cutover" item the person acts on when a moved send fails with recipient_suppressed. A call that adds one names the action before the deploy. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Re-import only now names bounces alone, and an app call that adds an address also records it where the moved sends can skip it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A moved send reaches every address an app's add call put on the provider's list, and an app-side flag misses past adds and survives a reactivation. So Step 3 asks once which sites rely on the call, and keeps them on the provider. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Step 3 asks which sites rely on the provider's list for addresses added by hand or by the app, and a non-answer keeps every site when the app adds addresses itself. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…the list The suppression-list question gets its own Step 3 section, asks only about sites that would move otherwise, keeps them on a non-answer, and says what would let them move. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…e work The question now comes right after the field checks, so a site it keeps needs no template copy or PostHog template, and the plan note carries what would let the site move. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
posthog-rs's async email send waits between retries with tokio::time::sleep and sends through reqwest, so a site on async-std or smol stays on Resend. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…o feat/migrate-email-brevo
Map Brevo's reason codes to PostHog's reason values, drop the per-sender and blocked-domain questions that the new Step 3 question replaces, strip the sync text from the blocked-list calls, and let a site that reads only the retry delay of a 429 move. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ender-scoped imports Read the blocked domains before Step 3's question when the key is set, count the sender-scoped entries imported as manual and the domains the moved sends reach, and take Brevo's documented x-sib-ratelimit-reset as its retry delay. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Step 3 asks about the suppression list before any template read, so offer the API read or the dashboard first, and name the sender column of the dashboard file the report counts from. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
There was a problem hiding this comment.
PostHog Review · Standard · reviewed 3f68103: 1 Should fix.
This is review round 10 on this pull request. When findings keep coming back in the same code, the approach is usually the problem. Before the next fix, check whether a different design would remove the whole class of findings.
#472's template check keeps a template the provider sent with no text part. A Brevo template has no text part of its own to turn off, and Brevo makes a plain text version of HTML emails itself, so the check doesn't keep it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
There was a problem hiding this comment.
PostHog Review · Standard · reviewed eca6619: 2 Should fix.
This is review round 11 on this pull request. When findings keep coming back in the same code, the approach is usually the problem. Before the next fix, check whether a different design would remove the whole class of findings.
… keep Node network failures without a response Brevo keeps an entry per sender, and PostHog keeps the first reason an address is imported with, so coalesce entries by address with bounce taking precedence. Set the Node error's response only when PostHog answered, so a catch that checks err.response still sees network failures as before. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ept across imports Count addresses, not entries, for the report, say that the bounce-first order holds within one import, and list the Node status names before the response condition. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Problem
A person who sends with Brevo and asks to move those sends to PostHog gets no help:
migrate-emailhas no Brevo variant.With this PR,
migrate-email-brevomoves Brevo sends to PostHog's email send in each language where both Brevo and PostHog have an SDK, andwizard migrate brevoruns it.Important
Stacked on #472, the core
migrate-emailskill. The base stays onfeat/migrate-email-all-sdksuntil #472 merges; then this PR lands on top. Each provider PR adds only its own provider and lands in any order: the lists they all touch,config.yaml,AGENTS.mdand the skill test's provider list, conflict in one line each as they land.It adds one provider folder, one config variant, one
AGENTS.mdrow and the provider in the bundle test. The shared steps it uses come from #472.Impact
Sends that arrive with
migratedFrom: 'brevo'show whether people use this skill to move off Brevo. Read it 14 days after the Wizard runsmigrateas a family.Caution
🔴 Merge risk: high · two-way door until the next context-mill release ships the skill; one-way once agents run it, because a migration's changes persist in each app's code and PostHog account
wizard migrate brevo, in every region. Nobody else changes: the shadow diff shows only the new Brevo bundle and its two index entries.brevo/account.md"What Step 3 asks about", "Map each entry" and "Blocked domains"): Brevo skips an unsubscribed or admin-blocked address for one sender, and every address on a blocked domain. Since feat(migrate-email): move Resend sends in every PostHog SDK's language #472, PostHog blocks a moved send only to an address that hard-bounced, and its list takes no domains. Every entry is imported with PostHog'sreason; a sender-scoped admin block or unsubscribe becomesmanual, which stops PostHog's marketing email to that address from every sender, and the report counts those addresses. feat(migrate-email): move Resend sends in every PostHog SDK's language #472's Step 3 question is the guard: Brevo's notes have it name the recipients' own unsubscribes and the blocked domains, read before the question from the API or the dashboard. Each site the person relies on Brevo's list for stays on Brevo; a site they don't name moves, and its sends reach those addresses and domains.Rollout
migratedFrom: 'brevo'and their error codes in Workflows; no alert pages on a wrong rewrite, since it shows up only in the person's app.Summary
references/brevo/holds the three files each variant has:sdk-reference.md: Brevo's SDKs in Node.js, Python, PHP, Ruby, Go, .NET and Java, including the earlier Sendinblue packages and each major version's client and error shape, and the HTTP API. It covers templates stored at Brevo, status reads, webhooks, and the APIs that are not sends.mapping.md:sendersaved byidor with a name and no address, more than oneto, areplyTowith a name,cc,bcc, attachments,headersother than the idempotency key,tags,scheduledAt,batchId,messageVersions, and a recipient's own tracking setting. An idempotency key moves only when it stays valid for PostHog and the code can't send it again for another email within 24 hours, since Brevo keeps a key for 15 minutes.captureEngagement. Anonymous tracking, per-contact consent, or an unknown answer about them keeps the site, open tracking off included, since a yes to PostHog's pixel doesn't cover how it ties opens to a person. Open tracking off otherwise follows feat(migrate-email): move Resend sends in every PostHog SDK's language #472's Step 3 ask: the site moves only when the person agrees to PostHog's open pixel, and an unknown open-tracking answer keeps it. Opens tracked per recipient moves, with a plan-row note on clicks, which PostHog counts only for a template made in PostHog that isn't marked sensitive. Where Brevo's click tracking was off or unknown, such a template must be marked sensitive, as feat(migrate-email): move Resend sends in every PostHog SDK's language #472's Step 3 says;account.mdnames the setting.template-to-code.mdsays: the code renders Brevo's template language with the site'sparamsand sends the renderedsubjectandhtml. A send with its own content andparamsis rendered the same way. Own content with{{or{%but noparams, or content the code doesn't hold in full, such as content from a database or a CMS, moves only when the person confirms from a sent email what Brevo did with it. A template made in PostHog is used only when the person asks for one, and one for an email with a login code or a reset link is marked sensitive. Every Brevo tag, filter, or placeholder whose output the docs don't pin, such as{{ contact.… }}without an app value,{{ unsubscribe }}, or HTML in a value, keeps the template on Brevo. A template Brevo lists as inactive, or whose status nobody can tell, keeps its site unless the person explicitly agrees that PostHog sends it, noted in the plan row. A Brevo template has no text part of its own, and Brevo's help center says Brevo makes a plain text version of HTML emails itself, so feat(migrate-email): move Resend sends in every PostHog SDK's language #472's check for a template sent with no text part doesn't keep it: the send hashtmlonly, and PostHog makes the text part.catchread for each SDK.400and422or their classes readvalidation_error; other classes read their status code; a check of the baseBrevoErrororHttpErrorreads afromPostHogmark. A429retry skipsquota_limited. A branch onBrevoTimeoutError, a 4.x or 5.xBrevoErrorbranch that goes on after the error, a legacy Pythonexcept ApiException, or a legacy PHPApiExceptioncatch with Guzzle 7, where a status-0 failure (such as TLS) took another path than other network failures, a Ruby branch on only one ofcode0(connection failure) or nocode(timeout), each of which tells one kind of network failure from another under feat(migrate-email): move Resend sends in every PostHog SDK's language #472's shared rule, a branch on Brevo's errorcodeor message text, and code that stores or returns the message id of a send PostHog could refuse as suppressed keep the site on Brevo. So do a read of the response headers of a Brevo send other than its retry delay (x-sib-ratelimit-reseton a429, which moves to PostHog's error), a Python or PHP send the Brevo SDK runs on a thread or a promise whose answer the code reads later, and a .NET branch onErrorCode0; any other branch on status0readsnetwork_error. Where the oldcatchreaderr.response.statusCode, the rethrown error setsresponseonly when PostHog answered, so a network failure still has none.timeoutInSecondsor anabortSignaltimeout on@getbrevo/brevo, PHP's'timeout', and a Go context deadline. The send gets no retries, as the core says. An axios orkytimeoutis a limit the PostHog SDK can't keep, so those sites stay, as the core says. Legacy Python's_request_timeout(urllib3'stotal, or connect and read apart) and Ruby'sconfig.timeout(Typhoeus's whole-request limit) are deadlines the PostHog SDK can't keep, so their sites stay, as the core's rules andrewrite-ruby.mdsay. Without a deadline, a retry count the code gives Brevo's client or send, such asmaxRetries: 0, carries over. A site whose Brevo SDK retried, such as408and5xxanswers in the newer SDKs or failed connections through urllib3 in legacy Python, gets a plan-row and report note on the retries it loses, since PostHog's SDK retries network failures and rate limits. A send a caller can cancel stays, unless the PostHog send takes the same cancellation, as in Go and .NET.$workflows_email_*event or to none. Aunique_openedhandler keeps only the first$workflows_email_openedof each send. A handler that matches events bymessage-id, a tag orX-Mailin-customkeeps its sites on Brevo, a handler of an event PostHog has no event for, such assoft_bounceordeferred, keeps them, and aclickhandler keeps them unless they move to a template made in PostHog that isn't marked sensitive.smtpBlacklistSender,blockNewDomain) and removes (the contact unblock,deleteHardbounces, the domain unblock), for feat(migrate-email): move Resend sends in every PostHog SDK's language #472's Step 3 question and Step 5's "After cutover" item. A domain unblock adds no item, since PostHog holds no domain. Names match in every SDK's casing.account.md: Brevo's code, DKIM and DMARC records next to PostHog's, and a new restricted API key for this migration with only Transactional email set to Read (a full-access key when the account has no restricted keys); the person deletes it after the suppression import, and a template copy needed later comes from the dashboard. It reads template content and the account's tracking settings, and imports each blocked address once with PostHog'sreason, from the dashboard file (by its reason labels) or the API:bouncewhen any of its entries is a hard bounce, elsecomplaintfor a spam complaint, elsemanualfor admin blocks and unsubscribes. Brevo keeps an entry per sender and PostHog keeps an address's first reason, so the order holds within one import, and the re-import tells the person that an address imported earlier ascomplaintormanualthat later hard-bounced stays that way. "What Step 3 asks about" names what feat(migrate-email): move Resend sends in every PostHog SDK's language #472's Step 3 question covers for Brevo. Its "For the report" section counts the addresses imported asmanualfor one sender's entries only and the blocked domains the moved sends reach.Evidence
ab778d1f:vitest run scriptspasses 344 tests in 30 files, the core's section-pointer check for Brevo's files included;node scripts/build.jsexits 0 and leaves the tree clean.migrate-email-skill.test.js):brevoin the provider list runs the shared per-provider checks: the Brevo bundle holds its three files and the eight rewrite files, every reference resolves, no other provider's folder ships with it, and it is amigratesubcommand.@getbrevo/brevo4.0.1 and 5.0.0 (BrevoTimeoutError extends Error), 6.2.0 (extends BrevoError, andBrevoErrorfor unknown network failures), 3.0.4 (HttpErroronly for answers; network errors reject with the raw request error), andbrevo-python1.2.0 andsib-api-v3-sdk1.1.4 (ApiException(status=0)on TLS failures, aPoolManagerwithoutretries, so urllib3's default retries apply, and_request_timeoutasurllib3.Timeout(total=…)for one number or connect and read for a pair), thebrevoRuby gem 4.0.0 (ApiErrorwithcode0 for libcurl failures and nocodefor a timeout, andconfig.timeoutas Typhoeus's limit on the whole request),getbrevo/brevo-php6.x (BrevoApiExceptionfor answers,BrevoExceptionfor network failures, andtimeoutandmaxRetriesoptions),@getbrevo/brevo6.2.0's fetcher (it retries only408,429and5xxanswers, never a request that timed out, and passes oneabortSignalto every attempt), and Guzzle 6.5 and 7.9 (ConnectExceptionunderRequestExceptionin 6; in 7 only connection errors areConnectException, the rest such as a certificate failure stayRequestException).getbrevosource read through raw.githubusercontent.com, after Fleet refused thegetbrevoorg. The Brevo attachments follow-up rechecks them against Brevo's API reference, OpenAPI spec and docs.x-sib-ratelimit-reset.090ac92b, feat(migrate-email): move Resend sends in every PostHog SDK's language #472's head) and at this head, unpacked and diffed. Onlymigrate-email-brevois new, and its 21 shared reference files are byte-identical tomigrate-email-resend's.manifest.jsonandskill-menu.jsonchange only by the build timestamp and the Brevo entries. Every other bundle, Resend's included, is unchanged.build.ymlnow runs, as this PR targetsmain, and passed on4b89b26f. The gap accepted while the base wasfeat/migrate-email-all-sdkscarries.🤖 Agent context
Decisions
captureEngagement: anonymous, per-contact consent or unknown stays; open tracking off moves only with the person's yes to PostHog's open pixel, as #472's Step 3 asks, and an unknown open-tracking answer stays; opens tracked per recipient moves. Replaced moving both off withcaptureEngagement: falsereason:bouncewhen any entry is a hard bounce, elsecomplaintfor a spam complaint, elsemanualfor admin blocks and unsubscribes. Blocked domains can't be imported and are counted. Sites that rely on Brevo's list for anything but hard bounces stay, through #472's Step 3 question. Replaced asking about sender-scoped entries and domains in Step 5, which #472's reason-based import made mootBrevoErrorbranch that goes on after the error, and a legacy PythonApiExceptionhandler that goes on after an unchecked error. Replaced moving them with a plan-row notevalidation_errorfor 400 and 422, the status code for the othersmapping.mdnames which Brevo options count. Without a deadline, the code's retry count carries over, and a site whose Brevo SDK retried answers notes the retries it loses. Replaced keeping it on BrevoparamsDisputed finding: #465 (comment), can't be checked: Brevo's docs don't say which failures
brevo_csharp'sApiExceptioncovers, and Fleet blocks thegetbrevoorg that holds its source. A .NET branch onErrorCode0keeps its site instead.Involvement: ReviewHog (2 threads fixed this pass); qa-swarm round 1 on
08d0a751(fresh-eyes, correctness, risk, design, anti-overengineering, sweep, and Team lenses for Michael and Analytics). Round 2 on488f076b(fresh-eyes, correctness, risk, design, anti-overengineering, team-michael), GPT-6.1 Sol final reads at091a3dc1,0ad04295,bf639d34(1 finding, fixed) anda1ce8e91(clean), and ReviewHog on091a3dc1(2 findings, fixed),0ad04295(1 finding, fixed) anda1ce8e91(clean). After the forward merge of #472 at8ead8e1b, a two-family final read oncb03d587(Opus 5.5 fresh eyes, GPT-6.1 Sol correctness: 5 findings, 4 fixed, 1 rejected), then onaf068f58(3 findings, fixed) and93767059(3 minor findings, fixed; correctness clean), the last past the six-read ceiling under Michael's standing review rule, and ReviewHog on93767059(1 should-fix, fixed), and02caea96(2 minor findings, fixed; correctness clean). After the second forward merge of #472 at7ccc0ee4, final reads on415031c3(1 finding, fixed) and975be2ef(clean in both families). After the third forward merge of #472 atf6583549, final reads on5995749c(1 nit, fixed) ande25eb563(skill text clean in both families). After the fourth forward merge of #472 at17c6ca15, a final read on2ecce7f6(correctness clean; fresh eyes 4 minor: 1 fixed, 1 rejected, 1 for the core's shared test, and this description), then one3f679be(clean in both families). The judge's risk line one3f679bfound the import left Brevo's sender and domain blocks out, so moved sends could reach those addresses; fixed in1d094244. After the fifth forward merge of #472 at090ac92b, which runs the section-pointer check for Brevo's files (the judge's review line),62700066says what a re-import does with Brevo's sender and domain blocks. Final read 16 on it (Opus 5.5 fresh eyes: 1 major and 3 minor; GPT-6.1 Sol correctness: 1 major), all fixed ina022e184: per-tenant senders, an unread list, the order of the domain and sender questions, and counts instead of addresses in the report. Final read 17 ona022e184(both families: a fixed-address site that keeps moving after a domain no skipped the sender check; fresh eyes also: the unread-list path lost the per-tenant rule), fixed in20e1be77. Final read 18 on20e1be77(correctness clean; fresh eyes 1 minor: an unread-list no kept only the sites named), fixed ineb08a7e4. Final read 19 oneb08a7e4(correctness clean; fresh eyes 2 minor on the unread-list wording and a missing no-answer outcome), fixed inbe25dbbb. Final read 20 onbe25dbbb(correctness clean; fresh eyes 1 minor: an unread-list "None" skipped the question for asmtpBlacklistSendercall), fixed inf9d38606. Final read 21 onf9d38606: clean in both families. ReviewHog onf9d38606(2 P2: Brevo keys can be restricted; ablockNewDomaincall blocks domains after the check), fixed infd7be30d. Final read 22 onfd7be30d(fresh eyes 3 minor: this description, the core's sync item for a domain block, the Pascal-case name; correctness 1 major: a fixed-address site kept moving with a domain-block call), fixed in085f02f1. Final read 23 on085f02f1(fresh eyes 2 minor: this description, the other calls' SDK casing), fixed ina3bd7dcc; final read 24 ona3bd7dccclean in both families. ReviewHog ona3bd7dcc(2 P2: a handler readingRetry-Afterfrom Brevo's error, .NET network failures), on141a1e26(2 P2: blocked contacts with no sender, legacy Python async sends) and onaae92a0c(2 P2:unique_openedhandlers,deleteHardbounces), all fixed but the .NET one, disputed below. Fleet blocks thegetbrevoGitHub org, so fromedd3b493on, this step cites only Brevo's API reference, OpenAPI spec and docs pages, and drops what only SDK source backed. Final reads 25 to 30 on141a1e26,aae92a0c,884dc501,f2ae8f37,a2ed7833and22c62242: correctness clean in each; fresh eyes found 2, 5, 3, 2, 2 and 1 (this description), fixed through22c62242. ReviewHog on22c62242(1 P2: an inactive Brevo template would become a live send), fixed in24c16710; final read 31 on it (correctness clean; fresh eyes: an unknown template status, and this description), fixed in57a6b7a8. Final read 32 on57a6b7a8: clean in both families, and ReviewHog on57a6b7a8: 0 findings;fleet proof gatepasses there. After #472 landed, main merged in at37f22b13, and4b89b26fadapts Brevo to its reason-based import. Final read 33 on4b89b26f(correctness clean; fresh eyes 1 major, this description, and 3 minor and 1 nit: the sender-scopedmanualentries and the domains go unreported, Step 3 asks before the domains are read, and Brevo's documented wait header isx-sib-ratelimit-reset), fixed in496de01f. Final read 34 on496de01f(correctness clean; fresh eyes 1 minor: Step 3 asks before any key is set, so the domains were never read; 1 nit: the dashboard file's sender column), fixed in3f68103a. Final read 35 on3f68103a: clean in both families, andfleet proof gatepasses there. ReviewHog on3f68103a(1 should-fix: #472's text-part check would keep every Brevo template), fixed ineca6619c. Final read 36 oneca6619c(correctness clean; fresh eyes 1 minor: this description), andfleet proof gatepasses there. ReviewHog oneca6619c(2 should-fix: an address with several Brevo entries could import asmanualbefore its hard bounce; the Node wrapper setresponseon network failures), fixed in11be57f3. Final read 37 on11be57f3(correctness clean; fresh eyes 3 minor and 1 nit: count addresses, not entries; the order holds within one import only; this description; the Node status list), fixed in55d01898. Final read 38 on55d01898(correctness clean; fresh eyes 1 minor: Step 5's counts are addresses, not Brevo entries; 1 nit: this description), fixed inab778d1f. Earlier passes on this branch ran 2 panel rounds and 17 final reads before the core skill moved templates into the code. Proof: the shadow diff ran; the live run, prod shape, and CI are gaps.Agent: Claude Code, Claude Opus 5.5 (
claude-opus-5-5, high), as the migration-skills slice of Silthus/posthog#461. Reviews on GPT-6.1 Sol (high), the fresh-eyes lens on Opus 5.5; fixes on Opus 5.5 (high).🤖 Generated with Claude Code