[{"data":1,"prerenderedAt":43},["ShallowReactive",2],{"doc-setup":3},{"html":4,"headings":5},"\u003Cp>HelpBites runs with nothing but Docker (\u003Ca href=\"\u002Fdocs\u002Fgetting-started\">Getting started\u003C\u002Fa>). Everything below is optional and\nindependent: switch on what you need, in any order. Each section says what you need, what to do, and how to tell it\nworked. The deep guides are linked from each, and every environment variable is in the\n\u003Ca href=\"\u002Fdocs\u002Fself-host#environment-reference\">reference\u003C\u002Fa>.\u003C\u002Fp>\n\u003Cp>Where a setting can be made in the panel or in \u003Ccode>.env\u003C\u002Fcode>, the panel is the easy way and \u003Ccode>.env\u003C\u002Fcode> is the way for a whole\nserver. After editing \u003Ccode>.env\u003C\u002Fcode>, restart with \u003Ccode>docker compose up -d\u003C\u002Fcode>.\u003C\u002Fp>\n\u003Cdiv class=\"table-wrap\">\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>\u003C\u002Fth>\n\u003Cth>Needs\u003C\u002Fth>\n\u003Cth>Without it\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#ai\">AI\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>An API key from an OpenAI-compatible provider\u003C\u002Ftd>\n\u003Ctd>No AI replies, drafts, titles or translations; the AI buttons answer &quot;no AI provider&quot;\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#email\">Email\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>A Mailgun account and a domain\u003C\u002Ftd>\n\u003Ctd>No mail to users; the send API answers \u003Ccode>503\u003C\u002Fcode>; the panel still works\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#inbound-email\">Inbound email\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>Mailgun\u003C\u002Ftd>\n\u003Ctd>Conversations start from the widget and API only\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#github\">GitHub\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>A GitHub App you register once\u003C\u002Ftd>\n\u003Ctd>No repo linking, commit-driven changelog or code-aware error notes\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#storage\">Storage\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>An S3-compatible bucket\u003C\u002Ftd>\n\u003Ctd>No image or file uploads (\u003Ccode>install.sh\u003C\u002Fcode> bundles one)\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#search-and-knowledge\">Search and knowledge\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>Nothing (bundled)\u003C\u002Ftd>\n\u003Ctd>Always on; the model is tunable\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#error-tracking\">Error tracking\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>GlitchTip or Bugsink\u003C\u002Ftd>\n\u003Ctd>No Errors page\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#revenue\">Revenue\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>A Stripe, App Store, RevenueCat, Paddle or Lemon Squeezy key\u003C\u002Ftd>\n\u003Ctd>No revenue beside conversations\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#custom-domains\">Custom domains\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>A DNS record per domain\u003C\u002Ftd>\n\u003Ctd>Public pages live on your HelpBites host\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#push-notifications\">Push notifications\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>An Apple developer account\u003C\u002Ftd>\n\u003Ctd>The native app works without push\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Ca href=\"#assistants-and-api-keys-mcp\">An AI assistant over MCP\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd>Nothing\u003C\u002Ftd>\n\u003Ctd>—\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\n\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2 id=\"ai\">AI\u003Ca class=\"heading-anchor\" href=\"#ai\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>HelpBites speaks the OpenAI chat-completions protocol, so any provider with that API works: OpenAI, Anthropic and\nGemini (through their compatibility endpoints), OpenRouter, Mistral, Groq, xAI, DeepSeek, or a model you run yourself\n(LiteLLM, Ollama, vLLM). You bring the key; the community edition never meters or limits it.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>In the panel (recommended).\u003C\u002Fstrong> \u003Cem>Settings → AI → Add a provider\u003C\u002Fem>:\u003C\u002Fp>\n\u003Col>\n\u003Cli>Pick the provider. The endpoint fills in; choose \u003Cstrong>Custom\u003C\u002Fstrong> for your own server and type its URL.\u003C\u002Fli>\n\u003Cli>Pick a model (type any model name your provider offers; the list is only suggestions).\u003C\u002Fli>\n\u003Cli>Paste your API key. It is stored sealed and never shown again.\u003C\u002Fli>\n\u003Cli>Optionally name an \u003Cstrong>Ultra\u003C\u002Fstrong> model, a stronger one used by the email designer's Ultra mode and other heavy jobs.\u003C\u002Fli>\n\u003Cli>Press \u003Cstrong>Test\u003C\u002Fstrong>. A reply from the model means it works; an error shows the provider's message.\u003C\u002Fli>\n\u003Cli>Save. With several providers, mark one active.\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Cp>A model on your own network (a LiteLLM or Ollama next to the app) needs \u003Ccode>ALLOW_PRIVATE_NETWORK=1\u003C\u002Fcode> in \u003Ccode>.env\u003C\u002Fcode>, because the\npanel refuses to point at private addresses by default.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>For the whole server.\u003C\u002Fstrong> Set \u003Ccode>LITELLM_API_BASE_URL\u003C\u002Fcode>, \u003Ccode>LITELLM_API_KEY\u003C\u002Fcode> and \u003Ccode>LLM_DEFAULT_MODEL\u003C\u002Fcode> in \u003Ccode>.env\u003C\u002Fcode>. Every\nworkspace then has an AI without adding a provider; a workspace's own provider still wins.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>Which one a call uses\u003C\u002Fstrong>, in order: the assistant's own endpoint (an assistant can bring its own endpoint, key and model, on its page under\n\u003Cem>Assistants\u003C\u002Fem>), then the workspace's active provider, then the server's gateway.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>What uses it\u003C\u002Fstrong>: AI assistants that answer customers (\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Felrumo\u002Fhelp-bites-oss\u002Fblob\u002Fmain\u002Fdocs\u002FKNOWLEDGE.md\" target=\"_blank\" rel=\"noopener\">Knowledge\u003C\u002Fa>), reply drafts in the inbox, the AI chat\nthat turns pasted text into feedback posts or planning items, titles for planning items, changelog entries drafted from\ncommits and shipped work, error analysis, the email designer, blog editing and translations of forms and pages.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>An assistant that replies by itself\u003C\u002Fstrong> is a separate switch, on purpose. \u003Cem>Assistants → New assistant\u003C\u002Fem>, give it\nknowledge, test it in the panel, and only then turn on auto-reply; its readiness list names what is still missing\n(a provider, knowledge to read, an answer it got right).\u003C\u002Fp>\n\u003Cp>Embeddings (how knowledge is searched) are separate and need no key: see \u003Ca href=\"#search-and-knowledge\">Search and knowledge\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2 id=\"email\">Email\u003Ca class=\"heading-anchor\" href=\"#email\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>HelpBites sends through \u003Ca href=\"https:\u002F\u002Fwww.mailgun.com\" target=\"_blank\" rel=\"noopener\">Mailgun\u003C\u002Fa>: replies to customers, notifications, verification codes,\nchangelog digests, campaigns and the transactional send API.\u003C\u002Fp>\n\u003Col>\n\u003Cli>\n\u003Cp>\u003Cstrong>Create a Mailgun account\u003C\u002Fstrong> and add a sending domain. Use a subdomain such as \u003Ccode>mail.yourdomain.com\u003C\u002Fcode> so your normal\nmail is untouched. Pick the EU or US region.\u003C\u002Fp>\n\u003C\u002Fli>\n\u003Cli>\n\u003Cp>\u003Cstrong>Add the DNS records\u003C\u002Fstrong> Mailgun shows (SPF, DKIM, and optionally the tracking CNAME) at your DNS host, and wait for\nthe domain to read \u003Cem>Verified\u003C\u002Fem>.\u003C\u002Fp>\n\u003C\u002Fli>\n\u003Cli>\n\u003Cp>\u003Cstrong>Create an API key\u003C\u002Fstrong> (Mailgun → your name → \u003Cem>API security\u003C\u002Fem> → \u003Cem>Add new key\u003C\u002Fem>).\u003C\u002Fp>\n\u003C\u002Fli>\n\u003Cli>\n\u003Cp>\u003Cstrong>Put it in \u003Ccode>.env\u003C\u002Fcode>:\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Cdiv class=\"code-block\">\u003Cbutton type=\"button\" class=\"copy-btn\" data-copy aria-label=\"Copy to clipboard\">Copy\u003C\u002Fbutton>\u003Cpre class=\"shj-lang-plain\">\u003Ccode>MAILGUN_API_KEY=key-…\nMAILGUN_DOMAIN=mail.yourdomain.com\nMAILGUN_FROM_EMAIL=support@mail.yourdomain.com   # on that domain, so SPF and DKIM line up\nMAILGUN_API_BASE_URL=https:\u002F\u002Fapi.mailgun.net      # US; the default is the EU https:\u002F\u002Fapi.eu.mailgun.net\u003C\u002Fcode>\u003C\u002Fpre>\u003C\u002Fdiv>\n\u003C\u002Fli>\n\u003Cli>\n\u003Cp>\u003Cstrong>Turn on delivery tracking\u003C\u002Fstrong> (so the email log learns about deliveries, opens, bounces and complaints, and bounces\nare suppressed automatically): Mailgun → \u003Cem>Sending\u003C\u002Fem> → \u003Cem>Webhooks\u003C\u002Fem> → copy the \u003Cstrong>HTTP webhook signing key\u003C\u002Fstrong> into\n\u003Ccode>MAILGUN_WEBHOOK_SIGNING_KEY\u003C\u002Fcode>, then add \u003Ccode>https:\u002F\u002FYOUR-HELPBITES\u002Fv1\u002Fwebhooks\u002Fmailgun\u003C\u002Fcode> for \u003Cem>Delivered, Opened, Clicked,\nPermanent failure, Temporary failure, Spam complaints\u003C\u002Fem> and \u003Cem>Unsubscribes\u003C\u002Fem>. Do this for every domain a project sends from.\u003C\u002Fp>\n\u003C\u002Fli>\n\u003Cli>\n\u003Cp>Restart, then open \u003Cstrong>Project → Email\u003C\u002Fstrong>. A strip across the top lists what is still missing and disappears when\nnothing is. Send yourself a test from a template, and check it in the \u003Cstrong>Log\u003C\u002Fstrong> tab.\u003C\u002Fp>\n\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Cp>Each project can send as itself (its own from name, address, sending domain and reply-to) under \u003Cem>Project → Email →\nSettings → Sender\u003C\u002Fem>. Campaigns and sequences also need a postal address there, the law in most places.\u003C\u002Fp>\n\u003Cp>Everything else (templates, variables, the designer, the API) is in \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Felrumo\u002Fhelp-bites-oss\u002Fblob\u002Fmain\u002Fdocs\u002FEMAIL.md\" target=\"_blank\" rel=\"noopener\">Email\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2 id=\"inbound-email\">Inbound email\u003Ca class=\"heading-anchor\" href=\"#inbound-email\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>Customers who write to \u003Ccode>support@yourdomain.com\u003C\u002Fcode> become conversations in the inbox, and your replies go back by email,\nthreaded. Needs Mailgun set up as above. Two ways, both in \u003Cem>Project → Email → Settings → Support email addresses\u003C\u002Fem>:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>Forward your mailbox.\u003C\u002Fstrong> HelpBites gives you an alias; your existing mailbox forwards to it. No DNS changes.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Use your domain.\u003C\u002Fstrong> Point a subdomain's MX at Mailgun. The panel shows the records.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>Either needs one Mailgun \u003Cem>Route\u003C\u002Fem> that forwards to HelpBites, and \u003Ccode>INBOUND_DOMAIN\u003C\u002Fcode> for the forwarding aliases. The\nsteps, with the values to paste, are in \u003Ca href=\"\u002Fdocs\u002Finbound-email\">Inbound email\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2 id=\"github\">GitHub\u003Ca class=\"heading-anchor\" href=\"#github\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>Connecting GitHub lets HelpBites \u003Cstrong>read\u003C\u002Fstrong> your repos (it never writes): AI error notes read the failing file,\n\u003Cem>Changelog → From commits\u003C\u002Fem> drafts entries from commit messages, the legal wizard suggests documents from your\ndependencies, and with the webhook on, \u003Ccode>#12\u003C\u002Fcode> in a commit or merged pull request links to board item 12 (and\n\u003Ccode>fixes #12\u003C\u002Fcode> moves it to done).\u003C\u002Fp>\n\u003Cp>There are two parts. The server owner registers a GitHub App once, then each workspace connects it.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>Once per server (about five minutes).\u003C\u002Fstrong> GitHub → \u003Cem>Settings → Developer settings → GitHub Apps → New GitHub App\u003C\u002Fem>:\u003C\u002Fp>\n\u003Col>\n\u003Cli>\n\u003Cp>Homepage URL is your \u003Ccode>APP_URL\u003C\u002Fcode>; Callback URL is \u003Ccode>APP_URL\u002Fv1\u002Fgithub\u002Fcallback\u003C\u002Fcode>; tick \u003Cem>Request user authorization\n(OAuth) during installation\u003C\u002Fem>.\u003C\u002Fp>\n\u003C\u002Fli>\n\u003Cli>\n\u003Cp>Optional but useful: tick \u003Cem>Active\u003C\u002Fem> on the webhook, set its URL to \u003Ccode>APP_URL\u002Fv1\u002Fwebhooks\u002Fgithub\u003C\u002Fcode> and a secret.\u003C\u002Fp>\n\u003C\u002Fli>\n\u003Cli>\n\u003Cp>Permissions: \u003Cem>Contents\u003C\u002Fem>, \u003Cem>Pull requests\u003C\u002Fem> and \u003Cem>Metadata\u003C\u002Fem>, all \u003Cstrong>read-only\u003C\u002Fstrong>. Subscribe to \u003Cem>Pull request\u003C\u002Fem> and \u003Cem>Push\u003C\u002Fem>.\u003C\u002Fp>\n\u003C\u002Fli>\n\u003Cli>\n\u003Cp>Create the app, copy the App ID and Client ID, generate a client secret and a private key (\u003Ccode>.pem\u003C\u002Fcode>).\u003C\u002Fp>\n\u003C\u002Fli>\n\u003Cli>\n\u003Cp>Put them in \u003Ccode>.env\u003C\u002Fcode>:\u003C\u002Fp>\n\u003Cdiv class=\"code-block\">\u003Cbutton type=\"button\" class=\"copy-btn\" data-copy aria-label=\"Copy to clipboard\">Copy\u003C\u002Fbutton>\u003Cpre class=\"shj-lang-plain\">\u003Ccode>GITHUB_APP_ID=…\nGITHUB_APP_SLUG=…\nGITHUB_APP_CLIENT_ID=…\nGITHUB_APP_CLIENT_SECRET=…\nGITHUB_APP_PRIVATE_KEY=…        # the .pem; literal \\n is accepted\nGITHUB_APP_WEBHOOK_SECRET=…     # if you turned the webhook on\nHELPBITES_SECRET=…              # signs the install link (openssl rand -base64 48)\u003C\u002Fcode>\u003C\u002Fpre>\u003C\u002Fdiv>\n\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Cp>The screenshots, every field and the upgrade notes are in\n\u003Ca href=\"\u002Fdocs\u002Fgithub#registering-the-github-app-server-owner-once\">GitHub\u003C\u002Fa>.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>Per workspace.\u003C\u002Fstrong> \u003Cem>Settings → GitHub → Connect GitHub\u003C\u002Fem>, install the app on your account or organisation and pick the\nrepos it may see. Then \u003Cem>Project → Settings → GitHub repos\u003C\u002Fem> picks the (up to 10) repos each project uses. Until the server\nis set up, the GitHub tab says &quot;GitHub isn't set up on this server&quot;.\u003C\u002Fp>\n\u003Cp>Using GitHub to publish \u003Cstrong>your own\u003C\u002Fstrong> changelog from commits, from a script or a CI job rather than the panel, is the\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Felrumo\u002Fhelp-bites-oss\u002Fblob\u002Fmain\u002Fcli\u002FREADME.md\" target=\"_blank\" rel=\"noopener\">command-line tool\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2 id=\"storage\">Storage\u003Ca class=\"heading-anchor\" href=\"#storage\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>Attachments, images in changelog entries and posts, brand logos and exports go to an S3-compatible bucket.\n\u003Ccode>scripts\u002Finstall.sh\u003C\u002Fcode> already configures the bundled \u003Ca href=\"https:\u002F\u002Frustfs.com\" target=\"_blank\" rel=\"noopener\">RustFS\u003C\u002Fa> one (\u003Ccode>COMPOSE_PROFILES=rustfs\u003C\u002Fcode>), so a\nfresh install works. For production, decide where files live:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>Keep the bundled bucket.\u003C\u002Fstrong> Back up its four volumes, and give \u003Ccode>sb-rustfs\u003C\u002Fcode> port \u003Ccode>9000\u003C\u002Fcode> a domain (the bundled Caddy\ndoes this as \u003Ccode>files.&lt;your domain&gt;\u003C\u002Fcode>). \u003Ccode>S3_PUBLIC_URL\u003C\u002Fcode> must be a URL browsers can reach.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Use your own\u003C\u002Fstrong> (AWS S3, Cloudflare R2, Backblaze, Garage, MinIO): set \u003Ccode>S3_ENDPOINT\u003C\u002Fcode>, \u003Ccode>S3_PORT\u003C\u002Fcode>, \u003Ccode>S3_USE_SSL\u003C\u002Fcode>,\n\u003Ccode>S3_ACCESS_KEY\u003C\u002Fcode>, \u003Ccode>S3_SECRET_KEY\u003C\u002Fcode>, \u003Ccode>S3_BUCKET\u003C\u002Fcode> and \u003Ccode>S3_PUBLIC_URL\u003C\u002Fcode>, and remove \u003Ccode>rustfs\u003C\u002Fcode> from \u003Ccode>COMPOSE_PROFILES\u003C\u002Fcode>.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>Check it by attaching an image to a feedback post. The settings for AWS and for the bundled bucket are in\n\u003Ca href=\"\u002Fdocs\u002Fself-host#s3-storage\">Self-hosting → S3 storage\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2 id=\"search-and-knowledge\">Search and knowledge\u003Ca class=\"heading-anchor\" href=\"#search-and-knowledge\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>Nothing to do: Meilisearch is bundled and the app creates its indexes on boot. Knowledge search is hybrid (words and\nmeaning); the meaning half runs an embedding model on the CPU inside the app. Three choices if that is not what you want:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Keep it (the default). First boot downloads the model, then it is cached on a volume.\u003C\u002Fli>\n\u003Cli>Offload it to an OpenAI-style embeddings service with \u003Ccode>EMBEDDING_ENDPOINT\u003C\u002Fcode> and \u003Ccode>EMBEDDING_API_KEY\u003C\u002Fcode>. Do this if you\nrun more than one app container.\u003C\u002Fli>\n\u003Cli>Turn it off with \u003Ccode>KNOWLEDGE_DISABLE_EMBEDDINGS=1\u003C\u002Fcode> for faster boots and plain keyword search.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>Before the app is public, set a strong \u003Ccode>MEILI_MASTER_KEY\u003C\u002Fcode>, then run \u003Ccode>docker compose exec app bun scripts\u002Fmeili-key.mjs\u003C\u002Fcode>\nand put the printed key in \u003Ccode>MEILI_API_KEY\u003C\u002Fcode>: the app then holds a key that can only do what it needs.\u003C\u002Fp>\n\u003Ch2 id=\"error-tracking\">Error tracking\u003Ca class=\"heading-anchor\" href=\"#error-tracking\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>Turn each project's errors into an \u003Cstrong>Errors\u003C\u002Fstrong> page with alerts and AI notes, through a \u003Ca href=\"https:\u002F\u002Fglitchtip.com\" target=\"_blank\" rel=\"noopener\">GlitchTip\u003C\u002Fa>\nor \u003Ca href=\"https:\u002F\u002Fwww.bugsink.com\" target=\"_blank\" rel=\"noopener\">Bugsink\u003C\u002Fa> you run next to HelpBites (both are Sentry-compatible, so your app uses its\nnormal Sentry SDK with a DSN HelpBites gives it).\u003C\u002Fp>\n\u003Col>\n\u003Cli>Start the tracker: \u003Ccode>COMPOSE_PROFILES=glitchtip\u003C\u002Fcode> or \u003Ccode>bugsink\u003C\u002Fcode> (add to the list, comma separated), set its \u003Ccode>*_SECRET_KEY\u003C\u002Fcode>\nand, for GlitchTip, \u003Ccode>GLITCHTIP_DB_PASSWORD\u003C\u002Fcode>, and give it a public domain (\u003Ccode>sb-glitchtip:8000\u003C\u002Fcode> or \u003Ccode>sb-bugsink:8000\u003C\u002Fcode>):\nyour app's SDK sends events there.\u003C\u002Fli>\n\u003Cli>Create an API token in the tracker and enter it under \u003Cem>Connect error tracking\u003C\u002Fem> in the panel (or in \u003Ccode>.env\u003C\u002Fcode>).\u003C\u002Fli>\n\u003Cli>Open \u003Cem>Project → Errors\u003C\u002Fem>, connect, and copy the DSN into your app's Sentry SDK.\u003C\u002Fli>\n\u003Cli>\u003Cem>Alerts\u003C\u002Fem> emails you on new or frequent errors (needs \u003Ca href=\"#email\">email\u003C\u002Fa>); \u003Cem>Analyze with AI\u003C\u002Fem> needs \u003Ca href=\"#ai\">AI\u003C\u002Fa>.\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Cp>The bootstrap steps for each tracker, one tracker project per app, and the SDK snippets are in \u003Ca href=\"\u002Fdocs\u002Ferrors\">Errors\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2 id=\"revenue\">Revenue\u003Ca class=\"heading-anchor\" href=\"#revenue\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>\u003Cem>Project → Revenue\u003C\u002Fem> shows MRR, plans, top customers and payments beside each conversation. Connect any mix of Stripe\n(a restricted read-only key), RevenueCat, the App Store (an App Store Connect API key, which also unlocks TestFlight\nnumbers), Paddle, Lemon Squeezy or your own API. Each source is a form in the panel that explains the key it needs and\ntests it: \u003Ca href=\"\u002Fdocs\u002Frevenue\">Revenue\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2 id=\"custom-domains\">Custom domains\u003Ca class=\"heading-anchor\" href=\"#custom-domains\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>Serve a project's hub, feedback board, roadmap, changelog and blog from your own host (\u003Ccode>feedback.example.com\u003C\u002Fcode>): add the\nhost in the project's settings, create the TXT and CNAME records the panel shows, and press verify. The bundled Caddy\ngets certificates only for verified hosts. With your own proxy you need the same two checks:\n\u003Ca href=\"\u002Fdocs\u002Fcustom-domains\">Custom domains\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2 id=\"push-notifications\">Push notifications\u003Ca class=\"heading-anchor\" href=\"#push-notifications\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>Only for the native iPhone, iPad and Mac app. Create an APNs auth key (\u003Cem>Apple Developer → Keys\u003C\u002Fem>), and set\n\u003Ccode>APNS_KEY_ID\u003C\u002Fcode>, \u003Ccode>APNS_TEAM_ID\u003C\u002Fcode>, \u003Ccode>APNS_BUNDLE_ID\u003C\u002Fcode>, \u003Ccode>APNS_KEY\u003C\u002Fcode> (the whole \u003Ccode>.p8\u003C\u002Fcode>) and \u003Ccode>APNS_ENV\u003C\u002Fcode> (\u003Ccode>sandbox\u003C\u002Fcode> for builds from\nXcode, \u003Ccode>production\u003C\u002Fcode> for TestFlight and the App Store). Building the app is in \u003Ca href=\"\u002Fdocs\u002Fdevelopment#the-native-app\">Development\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2 id=\"assistants-and-api-keys-mcp\">Assistants and API keys (MCP)\u003Ca class=\"heading-anchor\" href=\"#assistants-and-api-keys-mcp\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cp>Nothing to install on the server. In \u003Cem>Settings → API keys\u003C\u002Fem> you create an \u003Ccode>ak_…\u003C\u002Fcode> key for scripts, or you connect Claude,\nChatGPT, Cursor or any MCP client to \u003Ccode>https:\u002F\u002FYOUR-HELPBITES\u002Fmcp\u003C\u002Fcode> and sign in with your browser; the assistant then acts\nas you, in one workspace, and you can cut it off from the same page. \u003Ca href=\"\u002Fdocs\u002Fmcp\">MCP\u003C\u002Fa> lists what it can do.\u003C\u002Fp>\n\u003Ch2 id=\"is-it-working\">Is it working?\u003Ca class=\"heading-anchor\" href=\"#is-it-working\" aria-label=\"Link to this section\">#\u003C\u002Fa>\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Ccode>GET \u002Fv1\u002Fhealth\u003C\u002Fcode> reports each dependency (Parse, Redis, Meilisearch, the bucket, the worker).\u003C\u002Fli>\n\u003Cli>\u003Ccode>GET \u002Fv1\u002Fmeta\u003C\u002Fcode> reports the version, the edition and whether sign-up is open.\u003C\u002Fli>\n\u003Cli>\u003Cem>Project → Overview\u003C\u002Fem> carries the setup checklist, judged from what HelpBites has actually seen.\u003C\u002Fli>\n\u003Cli>Something off? \u003Ca href=\"\u002Fdocs\u002Fself-host#troubleshooting\">Self-hosting → Troubleshooting\u003C\u002Fa>.\u003C\u002Fli>\n\u003C\u002Ful>\n",[6,10,13,16,19,22,25,28,31,34,37,40],{"id":7,"text":8,"level":9},"ai","AI",2,{"id":11,"text":12,"level":9},"email","Email",{"id":14,"text":15,"level":9},"inbound-email","Inbound email",{"id":17,"text":18,"level":9},"github","GitHub",{"id":20,"text":21,"level":9},"storage","Storage",{"id":23,"text":24,"level":9},"search-and-knowledge","Search and knowledge",{"id":26,"text":27,"level":9},"error-tracking","Error tracking",{"id":29,"text":30,"level":9},"revenue","Revenue",{"id":32,"text":33,"level":9},"custom-domains","Custom domains",{"id":35,"text":36,"level":9},"push-notifications","Push notifications",{"id":38,"text":39,"level":9},"assistants-and-api-keys-mcp","Assistants and API keys (MCP)",{"id":41,"text":42,"level":9},"is-it-working","Is it working?",1794596525500]