n8n Engineering Daily Digest: Jira + GitHub to Slack

Scheduled trigger → parallel Jira and GitHub fetch → OpenAI summary → Slack Block Kit post → optional Google Sheets log

Who

Engineers, tech leads, or ops who need automated daily engineering updates for their team

You get

Receive a weekday morning Slack digest with Jira movement, GitHub PR status, and blockers; optionally sync to a spreadsheet

You need

n8n · google-sheets · github · slack

Steps

1. Configure trigger and global parameters

   After importing the template, set the 'When Weekday 9AM' Schedule Trigger with cron 0 9 * * 1-5 and the correct timezone. Then open the 'Set Config Fields' node and fill assignments: jiraBaseUrl (e.g. https://your-domain.atlassian.net), jiraProjectKey, staleTicketDays (default 3), staleReviewHours (default 48), repos (JSON array with owner and name per entry), slackChannel (e.g. #eng-daily-digest), sheetEnabled (true/false), sheetId, sheetTabName, openAiModel (default gpt-4o-mini), and optional userMap (Jira accountId / GitHub login → Slack userId).

2. Fetch and normalize Jira data in parallel

   The workflow splits into two Jira branches from Config: 1) 'Fetch Jira Recent Activity' uses JQL: project = {{ $json.jiraProjectKey }} AND updated >= -1d ORDER BY updated DESC, returning all results. 2) 'Fetch Jira Stale Tasks' uses JQL: project = {{ $json.jiraProjectKey }} AND status = "In Progress" AND updated <= -{{ $json.staleTicketDays }}d ORDER BY updated ASC. Downstream Code nodes normalize fields to {key, summary, status, assignee, assigneeAccountId, updated, url, daysStale?}, then 'Merge Jira Data' combines both streams into one Jira payload.

   Execution prompt · Source text。Normalize Jira Recent Data node (Code / JavaScript)

   Copy prompt

   const baseUrl = $('Set Config Fields').item.json.jiraBaseUrl;
   const issues = $input.all().map(item => {
     const issue = item.json;
     return {
       key: issue.key,
       summary: issue.fields?.summary,
       status: issue.fields?.status?.name,
       assignee: issue.fields?.assignee?.displayName || 'Unassigned',
       assigneeAccountId: issue.fields?.assignee?.accountId || null,
       updated: issue.fields?.updated,
       url: `${baseUrl}/browse/${issue.key}`
     };
   });
   return [{ json: { recentIssues: issues } }];

   Execution prompt · Source text。Normalize Jira Stale Data node (Code / JavaScript)

   Copy prompt

   const baseUrl = $('Set Config Fields').item.json.jiraBaseUrl;
   const now = Date.now();
   const issues = $input.all().map(item => {
     const issue = item.json;
     const updated = new Date(issue.fields?.updated).getTime();
     const daysStale = Math.floor((now - updated) / 86400000);
     return {
       key: issue.key,
       summary: issue.fields?.summary,
       status: issue.fields?.status?.name,
       assignee: issue.fields?.assignee?.displayName || 'Unassigned',
       assigneeAccountId: issue.fields?.assignee?.accountId || null,
       updated: issue.fields?.updated,
       daysStale,
       url: `${baseUrl}/browse/${issue.key}`
     };
   });
   return [{ json: { staleIssues: issues } }];

3. Fetch and normalize GitHub data in parallel

   The 'Split Repository List' Code node fans out Config.repos into one item per repo and computes sinceISO (24h ago) and staleDateISO. Then two branches run in parallel: 1) 'Fetch GitHub PR Activity' uses the native GitHub node repository.getPullRequests(state=all, sort=updated, direction=desc), then 'Flatten GitHub PR Activity' expands arrays, and 'Normalize GitHub Recent Data' classifies PRs into prsOpened / prsMerged / prsClosed against a 24h cutoff. 2) 'Fetch Stale GitHub PRs' uses an HTTP Request to GET https://api.github.com/search/issues?q=repo:{owner}/{name} is:pr is:open review:none&per_page=50, then 'Flatten GitHub Stale PRs' extracts body.items, and 'Normalize GitHub Stale Data' filters by staleReviewHours and computes hoursOpen. Both streams are combined by 'Merge GitHub Data'.

   Execution prompt · Source text。Split Repository List node (Code / JavaScript)

   Copy prompt

   const config = $json;
   const sinceISO = new Date(Date.now() - 24*60*60*1000).toISOString();
   const staleHours = config.staleReviewHours || 48;
   const staleDateISO = new Date(Date.now() - staleHours*60*60*1000).toISOString().split('T')[0];
   return (config.repos || []).map(repo => ({
     json: {
       owner: repo.owner,
       name: repo.name,
       fullName: `${repo.owner}/${repo.name}`,
       sinceISO,
       staleDateISO
     }
   }));

   Execution prompt · Source text。Flatten GitHub PR Activity node (Code / JavaScript)

   Copy prompt

   const results = [];
   for (const item of $input.all()) {
     const body = item.json;
     if (Array.isArray(body)) {
       for (const pr of body) {
         if (pr && pr.number != null) results.push({ json: pr });
       }
     } else if (body && body.number != null) {
       results.push({ json: body });
     }
   }
   return results.length ? results : [{ json: { number: null } }];

   Execution prompt · Source text。Normalize GitHub Recent Data node (Code / JavaScript)

   Copy prompt

   const now = Date.now();
   const cutoff = now - 24*60*60*1000;
   const prs = $input.all().map(i => i.json).filter(pr => pr && pr.number != null);
   const opened = [], merged = [], closed = [];
   for (const pr of prs) {
     const createdAt = new Date(pr.created_at).getTime();
     const mergedAt = pr.merged_at ? new Date(pr.merged_at).getTime() : null;
     const closedAt = pr.closed_at ? new Date(pr.closed_at).getTime() : null;
     const entry = {
       number: pr.number,
       title: pr.title,
       author: pr.user?.login,
       url: pr.html_url,
       repo: pr.base?.repo?.full_name
     };
     if (createdAt >= cutoff) opened.push(entry);
     if (mergedAt && mergedAt >= cutoff) merged.push(entry);
     if (closedAt && !mergedAt && closedAt >= cutoff) closed.push(entry);
   }
   return [{ json: { prsOpened: opened, prsMerged: merged, prsClosed: closed } }];

   Execution prompt · Source text。Flatten GitHub Stale PRs node (Code / JavaScript)

   Copy prompt

   const results = [];
   for (const item of $input.all()) {
     const body = item.json;
     const list = Array.isArray(body?.items) ? body.items : [];
     for (const issue of list) {
       if (issue && issue.number != null) results.push({ json: issue });
     }
   }
   return results.length ? results : [{ json: { number: null } }];

   Execution prompt · Source text。Normalize GitHub Stale Data node (Code / JavaScript)

   Copy prompt

   const staleReviewHours = $('Set Config Fields').item.json.staleReviewHours || 48;
   const cutoffMs = staleReviewHours * 60 * 60 * 1000;
   const now = Date.now();
   const prs = $input.all().map(i => i.json).filter(pr => pr && pr.number != null);
   const stale = prs
     .filter(pr => (now - new Date(pr.created_at).getTime()) > cutoffMs)
     .map(pr => ({
       number: pr.number,
       title: pr.title,
       author: pr.user?.login,
       url: pr.html_url,
       repo: (pr.repository_url || '').split('/repos/')[1],
       hoursOpen: Math.floor((now - new Date(pr.created_at).getTime()) / 3600000)
     }));
   return [{ json: { prsAwaitingReview: stale } }];

4. Summarize with OpenAI and parse

   'Merge Overall Digest' combines the Jira and GitHub payloads, then passes them to 'Summarize Digest with OpenAI'. The node uses native OpenAI Chat Complete with a system prompt requiring strict JSON output (no markdown, no code fences) containing whatMoved / whatShipped / whatIsBlocked / needsReview / sprintHealth. The user message serializes the merged data as JSON. Parameters: chatModel reads Config.openAiModel (default gpt-4o-mini), maxTokens=2000, temperature=0.2. Downstream 'Parse LLM Summary' parses choices[0].message.content; on failure it returns a degraded summary with error annotations.

   Execution prompt · Source text。Summarize Digest with OpenAI node system message

   Copy prompt

   You are an engineering-ops assistant. You will be given JSON describing the last 24 hours of Jira and GitHub activity for a software team. Produce a concise daily digest as STRICT JSON only, no markdown, no code fences, matching exactly this shape:
   {
     "whatMoved": "string, 2-4 sentences",
     "whatShipped": "string, 2-4 sentences",
     "whatIsBlocked": "string, 2-4 sentences, mention specific ticket keys",
     "needsReview": "string, 2-4 sentences, mention specific PR numbers/titles",
     "sprintHealth": "one sentence overall read, e.g. On track / At risk / Needs attention, with a short reason"
   }
   Only use the data given. If a category is empty, say so briefly. Do not invent tickets or PRs.

   Execution prompt · Source text。Summarize Digest with OpenAI node user message (expression, evaluated at runtime)

   Copy prompt

   ={{ JSON.stringify({ staleTicketThresholdDays: $('Set Config Fields').item.json.staleTicketDays, staleReviewThresholdHours: $('Set Config Fields').item.json.staleReviewHours, recentIssues: $json.recentIssues, staleIssues: $json.staleIssues, prsOpened: $json.prsOpened, prsMerged: $json.prsMerged, prsClosed: $json.prsClosed, prsAwaitingReview: $json.prsAwaitingReview }) }}

   Execution prompt · Source text。Parse LLM Summary node (Code / JavaScript)

   Copy prompt

   const response = $json;
   const digest = $('Merge Overall Digest').item.json;
   let parsed;
   try {
     const content = response.choices[0].message.content;
     parsed = JSON.parse(content);
   } catch (e) {
     parsed = {
       whatMoved: 'Summary unavailable — the LLM response could not be parsed.',
       whatShipped: '',
       whatIsBlocked: '',
       needsReview: '',
       sprintHealth: 'Unknown — check the digest workflow run for errors.'
     };
   }
   return [{ json: { summary: parsed, digest } }];

5. Build and send Slack Block Kit message

   The 'Build Slack Message Blocks' Code node reads Config.userMap and maps Jira assigneeAccountId / assignee and GitHub author to Slack @mentions (<@USER_ID>). It then assembles Block Kit JSON: a header (📋 Engineering Daily Digest), sprintHealth section, divider, whatMoved / whatShipped / whatIsBlocked / needsReview sections, divider, stale tickets list, PRs awaiting review list, and an optional context block with cc @mentions. 'Post Digest to Slack' sends it with messageType=block to the channel matching Config.slackChannel.

   Execution prompt · Source text。Build Slack Message Blocks node (Code / JavaScript)

   Copy prompt

   const { summary, digest } = $json;
   const userMap = $('Set Config Fields').item.json.userMap || {};
   
   function mention(idOrLogin) {
     const slackId = userMap[idOrLogin];
     return slackId ? `<@${slackId}>` : (idOrLogin || 'unassigned');
   }
   
   const staleMentions = (digest.staleIssues || []).map(i => mention(i.assigneeAccountId || i.assignee));
   const reviewMentions = (digest.prsAwaitingReview || []).map(p => mention(p.author));
   const allMentions = [...new Set([...staleMentions, ...reviewMentions])].filter(m => m.startsWith('<@'));
   
   const blocks = [
     { type: 'header', text: { type: 'plain_text', text: '📋 Engineering Daily Digest' } },
     { type: 'section', text: { type: 'mrkdwn', text: `*Sprint health:* ${summary.sprintHealth}` } },
     { type: 'divider' },
     { type: 'section', text: { type: 'mrkdwn', text: `*What moved*\n${summary.whatMoved}` } },
     { type: 'section', text: { type: 'mrkdwn', text: `*What shipped*\n${summary.whatShipped}` } },
     { type: 'section', text: { type: 'mrkdwn', text: `*What's blocked*\n${summary.whatIsBlocked}` } },
     { type: 'section', text: { type: 'mrkdwn', text: `*Needs review*\n${summary.needsReview}` } },
     { type: 'divider' },
     { type: 'section', text: { type: 'mrkdwn', text: `*Stale tickets (${(digest.staleIssues||[]).length})*\n` + ((digest.staleIssues||[]).map(i => `• <${i.url}|${i.key}> — ${i.summary} (${i.daysStale}d, ${mention(i.assigneeAccountId || i.assignee)})`).join('\n') || '_none_') } },
     { type: 'section', text: { type: 'mrkdwn', text: `*PRs awaiting review (${(digest.prsAwaitingReview||[]).length})*\n` + ((digest.prsAwaitingReview||[]).map(p => `• <${p.url}|#${p.number} ${p.title}> — ${p.hoursOpen}h, ${mention(p.author)}`).join('\n') || '_none_') } }
   ];
   
   if (allMentions.length) {
     blocks.push({ type: 'context', elements: [{ type: 'mrkdwn', text: `cc ${allMentions.join(' ')}` }] });
   }
   
   return [{ json: { blocks, text: `Engineering Daily Digest — ${summary.sprintHealth}`, digest, summary } }];

6. Optional Google Sheets logging

   'If Logging to Sheet Enabled' checks whether Config.sheetEnabled is true. If true, 'Append Row to Google Sheets' appends a row in autoMapInputData mode with fields ok, channel, message, message_timestamp; the target sheet must already contain matching column names. If false, the 'Skip Sheet Logging' noOp branch ends the workflow.

7. Credentials, invites, and optional user mapping

   1) Add n8n credentials for Jira Software Cloud, GitHub, OpenAI, and Slack (plus Google Sheets if sheetEnabled is true). 2) The GitHub token needs repo and pull-request read scope to support the review:none search. 3) Invite the Slack app/bot to the target channel and confirm it has permission to post messages. 4) Optionally populate Config.userMap with Jira accountId and GitHub login to Slack member ID mappings to enable @mentions in the digest.

You are setting up an n8n engineering daily digest workflow. Confirm in this order:
1) Check that your n8n instance supports the node versions used (scheduleTrigger v1.2, set v3.4, jira v1, github v1.1, openAi v1.1, slack v2.3, googleSheets v4.5, httpRequest v4.2, merge v3, if v2, code v2).
2) Ensure all required credentials are added: Jira Cloud, GitHub, OpenAI, Slack, and optionally Google Sheets.
3) After importing workflow.json, update Set Config Fields with your actual jiraBaseUrl, jiraProjectKey, repos, slackChannel, etc. Note that repos is a JSON array.
4) Verify the Schedule Trigger timezone matches your team's location.
5) Run manually once before enabling the schedule to verify OpenAI returns parseable JSON; if parsing fails, review the degraded output in Parse LLM Summary and adjust the system prompt or maxTokens.
6) If sheet logging is enabled, create the target sheet columns (ok, channel, message, message_timestamp) before running.