AfterLaunch: the agentic growth marketing engine

Growth content experiments already run, newest first, with scoped evidence. Read-only; never runs one. Each is an observation, not a rate.

One Memory page by slug, with its full markdown body: the record the growth marketing engine drafts from. Read-only, free.

The automated growth marketing loops provisioned for this product (AI visibility, competitors, channel distribution) with their autonomy level, plus runnable_loop_types: what runs on demand. A runnable type only runs once a loop of that type is provisioned and enabled here.

One growth move: evidence, source material, instructions and prepared_result. linked_article is earlier work to reuse, not a completed revision. Ready means awaiting review, not published. Closed post-trial access withholds bodies without asserting absence. tracked_link_state explains a null link (no_link, unavailable, withheld); null clicks means unreadable. Read-only, free.

What the shipped growth moves earned: tracked-link clicks per move, shipped counters, the weekly movement scoreboard, and whether GA4 and Search Console are connected. Read-only, free.

Read a queue draft by id. SEO linked_article {id,title,content} adds the full Markdown article.

Did the growth work move anything? The proof read for a period: AI visibility (GEO) at the START and the END of the window, the moves shipped between them, the scans that ran, and the receipts with what was verified live and the clicks each earned. ALWAYS read like_for_like before reporting a change: true means both scans asked the same tracked questions so the movement is real, false means the questions or answer source changed in the window, so it is a new baseline rather than a rise or a fall, and null means it cannot be proven either way. Never present a delta as a result when like_for_lik…

The site-wide SEO audit: issue groups with severity and affected pages, the weekly score trend, page performance, links and authority, and day-0 search ranks. Read before any SEO work of your own; the Feed seo_fix moves are minted from it. Read-only, free. Pass include_pages for the per-page detail.

The paste-able 'Growth (AfterLaunch)' section for this repository's CLAUDE.md or AGENTS.md: how a coding agent should use AfterLaunch for growth, marketing, SEO, AI visibility (GEO), competitor and distribution work, and which tools cost money. Free, read-only. Show it to the founder and add it with their say-so.

The Growth Snapshot this product run started from: positioning, the discoverability score, the actions prioritised by impact, the competitor set, and the day-0 SEO and AI visibility baselines. The stable business context, not live signal data. Read-only, free.

The daily growth marketing standup in one call: what recently shipped and what AfterLaunch verified about it (receipts), the top ranked moves shippable now (ready; rank is the position among all moves, so gaps mean those ranks are not shippable now), what the engine did on its own in the last 7 days (while_you_were_gone; capped at the account's age), and whether the deep scan is still running. Relay the message field. Read-only, free.

The measured AI visibility (GEO) results: how ChatGPT, Gemini, Perplexity and Google AI Overviews answer the tracked customer questions, plus share of voice, cited sources, competitor-owned gaps, the trend, AI crawlability and the off-site reach fold. Every reading carries its own sample count, and the sampling block says what such a count licenses, so treat a change as indicative unless it says otherwise. Read-only, free. Pass prompt_id to drill into one question.

The structured voice every AfterLaunch growth marketing draft is written in: tone, characteristic phrases, vocabulary, sentence style and what it avoids. register says which you got, and they do not sound alike: brand is the website's voice, personal is the founder's own writing, imported came from a brief. has_profile false means that voice is not extracted yet: an answer, not an error. Read-only, free.

Which growth data sources and distribution channels are connected: GA4 and Search Console (the measurement behind SEO and outcomes), Google Business Profile, and the LinkedIn / Reddit posting channels. Each row carries an honest state: connected, not_connected, or not_available with the reason it is shut on this account. Read-only, free.

The ranked backlog of growth marketing moves prepared for this product across SEO, AI visibility (GEO), competitors and distribution: title, why, area, status, rank, whether readable full-draft material is waiting (has_draft), and draft_kind (a full draft, a brief to complete, or null). It also carries posture: who may do it, act (you can), draft_only (you draft, a person posts) or founder_voice (the founder only). Read it before improvising growth work of your own. The material BODY is not here. Live moves only unless you pass include_resolved; pass since for what is new while the deep scan a…

The Memory pages: the shared record the growth marketing engine drafts from (business basics, audience, competitors, voice samples, recent observations). Summaries with a size hint. Read-only, free.

The drafted marketing and distribution content produced for this product, newest first: id, loop_type, kind, status and title. Read-only, free.

Unclaimed preparation jobs, oldest first. Each row carries job_id and its nullable attempt_id. Queue summaries never include prepared_content. Free; requires the 'act' scope.

The customer questions AfterLaunch tracks: the AI visibility (GEO) question set every measurement is taken against. Returns prompts in tracking order, cap (a hard 15) and used, plus curated: false means AfterLaunch generated them and any can be replaced. Read-only, free, on every plan.

Try to claim one queued preparation attempt. Pass job_id and the exact nullable attempt_id of a queued job. Start work only when the response says claim_won:true. An unchanged result is a non-winner and must not start work. Free; requires the 'act' scope.

Did the founder finish connecting the growth data source you handed them a link for? Pass the poll_token you got with that link. Reports pending (wait poll_after_seconds and ask again), connected, expired (mint a fresh link) or not_found. Idempotent: nothing is spent or consumed. Read-only, free.

Add a pending move to your user's growth moves. WHEN: you found a thread on a community they work, or their plan calls for a placement ask, outreach, a directory listing or a page for their site, and no move covers it yet. A repeat (same kind and target, or kind and title) hands back the existing move id. It comes back with its posture: a Reddit or Hacker News target is draft_only, meaning you draft and your user posts, and you never post there yourself without their yes in your thread. Outreach goes from their address, never without their yes. Free; 'act' scope; a fixed number a day. WHAT COM…

Put one measured number the founder will stand behind in public into the claim library, so every later reply and growth marketing post can cite it with its caveat. WHEN: a real measurement lands and the founder says they would defend it. Record the NUMBER and the caveat together; a number without its caveat is what this library exists to prevent. Never record an estimate, a projection, a rival's figure, or anything you inferred rather than measured. Free; requires the 'act' scope. Recording the same claim twice replaces your earlier wording; a pinned claim is left as it is. WHAT COMES BACK: th…

File a growth idea (optional channels, kind, score, final_text sent verbatim, desk: a post, reply or task the owner posts by hand, queued at slot_at, or a decision or read that needs the owner, due_on a date), original X draft or research outcome. Research appends a result to its idea; send expected_updated_at unchanged from the calendar. It keeps evidence holds and never authorises drafting. Requires 'act' scope; no spend or posting. Keep run_key stable. Drafts need action_kind social_post and one idea_id or source_idea; source_url is evidence. adopt_legacy only reclassifies an exact saved re…

Write one durable thing you have learned about this founder or their product into AfterLaunch's Memory, the record every growth and marketing draft is written from. WHEN: the moment the founder tells you something that will still be true next month and that AfterLaunch could not have measured itself. A standing preference, a rule about how they write, a fact not on the site yet, or what a shipped move actually earned. Recording it is part of the job, not a favour: AfterLaunch drafts from Memory, so anything you keep to yourself is a correction the next draft will need again. Record what LASTS.…

File one public profile or listing the person already has on a directory site, so AfterLaunch counts it as a verified presence and a weekly check confirms it is still live. WHEN: you opened the page and it is theirs. One address per site; a new address replaces the old one, the same address again changes nothing. The address must sit on that site's own host. Free; requires the 'act' scope. WHAT COMES BACK: every profile on record, the new one marked.

Save one writing rule the person approved into their OWN voice, never the brand's. The morning marketing posts drafted for them follow it, move and voice reads carry it, and they see and remove it in Memory. WHEN: only after they approve it in words, e.g. a yes to a rule you proposed from how they edited a draft before posting. Never record one they did not approve. One sentence. At most 12. Free; requires the 'act' scope.

Replace a drafted marketing output's body before it is shipped. Manual replacement only: the AI redraft stays in the app. A draft carrying an unfilled placeholder or a broken link is refused, and the previous draft is kept as revision history. Free; requires the 'act' scope and Founder tier.

SPENDS MONEY: run an on-demand AI visibility scan (scan='ai_visibility'), the GEO measurement of how the answer engines represent this product against its competitors. Needs the 'write' scope, a paid plan or an active trial, and an Idempotency-Key you mint. Bounded by the credit balance and the per-tenant daily and monthly caps. One scan per type per product per local day; a repeat that day replays the original result, regardless of the key.

SPENDS MONEY: run the growth-move generator on demand to top up the Feed with fresh marketing moves across SEO, AI visibility, competitors and distribution. Needs the 'write' scope, a paid plan or an active trial, and an Idempotency-Key you mint. Bounded by the credit balance and the per-tenant daily and monthly caps. Once a day per product.

Report a claimed attempt with its exact attempt_id: running; needs_input with one question for your user; shipped with the live published_url, or planned_url once merged (we check it); ready_for_review with prepared_content, or asset_url when your user must post it or when it is the pull request you opened; failed with fail_reason. Ready_for_review can be reported again, improved, until your user uses it. If you cannot read the whole thread, say so and ask your user for a screenshot. Free; requires the 'act' scope.

Apply your user's explicit decision to revise an accepted growth move's current managed attempt: its prepared or failed result is set aside and a new preparation attempt starts, only if this wins. Pass the exact job_id and expected_attempt_id. This never publishes anything. Free; requires the 'act' scope.

Resolve one drafted marketing output in the founder's review queue: action 'ship' marks it done or posted WITHOUT publishing anywhere, action 'skip' dismisses it. WHEN: call 'ship' the moment the founder confirms the content is actually out, with their go and never on your own; call 'skip' when they decide against it. Posting is the human's decision; reporting the outcome is yours, and it is not optional, because an unreported result leaves the record blind to what the work earned. Report only what is true: a draft you have written is not a draft that went out. This never posts to a channel: r…

Write the calibrated leans from a voice round into the FOUNDER'S OWN voice profile, the personal register, never the brand's. Two voices live here and they do not sound alike: the brand's came from the website and is what marketing pages and listings use, and this verb cannot touch it. WHEN: after a round, once an axis has enough answers to read. Send one write per round, not one per pair, and leave out an axis with too few observations: that is not a lean. Send only the axes that moved; the rest keep what they had. Free; requires the 'act' scope. WHAT COMES BACK: every axis on record, and the…

Record completion only after the founder explicitly confirms it. Use prepared_result after reviewing the current prepared attempt, or independent_work when the founder completed it separately. Site changes need a live published_url to ship; with planned_url they stay prepared. This records an attestation; it never publishes anything. Free; requires the 'act' scope.

Dismiss a growth move (Skip in the Feed). Free; requires the 'act' scope. Optional feedback_key (closed vocabulary, see the enum) plus a free-text feedback_note explain why, which improves future marketing drafts.

Connect a growth data source or distribution channel: mint a one-time link the FOUNDER opens in their own browser to approve it. You never complete the consent yourself; it is bound to their session. Returns url (hand it to the founder), poll_token, poll_after_seconds and expires_at; the link lasts 15 minutes. Refuses in plain words, before minting anything, when a connector is shut on this account. Free, spends nothing.

Add one customer question to the AI visibility (GEO) set AfterLaunch measures this product on. Free, spends nothing. Two honest consequences: changing the set makes the NEXT scheduled scan run at full depth rather than skipping ahead, and it re-baselines the week-over-week trend, because a comparison across two different question sets is not a real move. So add deliberately rather than churning the list. 10 to 200 characters, deduplicated, up to the included 50. Requires the 'config' scope and the Founder plan; on a free trial the questions stay readable.

Restore a just-shipped or just-skipped growth move to pending: the self-correction verb for a wrong ship or skip. Free; requires the 'act' scope. Idempotent on a move already pending. It never unwinds a real channel publish: a posted output stays posted, and an archived or expired move cannot be restored.

Remove one customer question from the AI visibility (GEO) set, freeing a slot against the included cap of 50. Free, spends nothing. Same two honest consequences as adding one: the next scheduled scan runs at full depth, and the week-over-week trend re-baselines. Matched on the question text, ignoring case and punctuation, and IDEMPOTENT. Requires the 'config' scope and the Founder plan.

Replace a pending growth move's marketing draft body before shipping. A draft carrying an unfilled placeholder or a broken link is refused rather than saved. Free; requires the 'act' scope.

Confirm the AfterLaunch key works: which product run it is bound to, its scopes, tier and trial_ends_at, plus onboarding (feed_ready is false while the Growth Snapshot deep scan is still building) and credits once metering is live.

Apply your user's explicit decision to cancel (withdraw) unfinished work on an accepted growth move: the job is withdrawn and the move returns to pending. Pass the exact current job and attempt; legacy work also needs the exact move and legacy job states and updated_at. This never publishes anything. Free; requires the 'act' scope.