{ "name": "Tech_Stack_Directional_Signal_Agent", "nodes": [ { "parameters": { "formTitle": "Tech Stack Directional Signal Agent", "formDescription": "Enter a company's GitHub details to generate an engineering direction intelligence report.", "formFields": { "values": [ { "fieldLabel": "Company Name", "placeholder": "e.g. OpenAI", "requiredField": true }, { "fieldLabel": "GitHub Org Slug", "placeholder": "e.g. openai (the part after github.com/)", "requiredField": true }, { "fieldLabel": "Sector", "placeholder": "e.g. Foundation Models, Fintech, DevTools" }, { "fieldLabel": "Stock Ticker", "placeholder": "e.g. MSFT (leave blank if private)" } ] }, "responseMode": "lastNode", "options": { "respondWithOptions": { "values": { "formSubmittedText": "✅ Report generated!" } } } }, "id": "6f71ff09-fad4-4a1c-8f2e-785b33307144", "name": "Form Trigger", "type": "n8n-nodes-base.formTrigger", "typeVersion": 2.2, "position": [ -528, -96 ] }, { "parameters": { "jsCode": "const form = $input.first().json;\n\nconst config = {\n recent_months: 6,\n lookback_months: 18,\n lang_delta_pct: 15,\n burst_multiplier: 2.0,\n min_stars: 10,\n velocity_repos: 5,\n velocity_z_thresh: 2.0,\n};\n\nconst org = (form['GitHub Org Slug'] || '').trim().toLowerCase();\nconst name = (form['Company Name'] || '').trim();\n\nif (!org || !name) {\n throw new Error('Company Name and GitHub Org Slug are required.');\n}\n\nconst graphql_query =\n '{ organization(login: \"' + org + '\") {' +\n ' repositories(first: 100, orderBy: {field: PUSHED_AT, direction: DESC}, privacy: PUBLIC) {' +\n ' nodes {' +\n ' name nameWithOwner stargazerCount forkCount isFork' +\n ' createdAt pushedAt' +\n ' primaryLanguage { name }' +\n ' languages(first: 15, orderBy: {field: SIZE, direction: DESC}) {' +\n ' edges { size node { name } } totalSize' +\n ' }' +\n ' repositoryTopics(first: 10) { nodes { topic { name } } }' +\n ' }' +\n ' }' +\n '} }';\n\nreturn [{\n json: {\n name,\n org,\n sector: (form['Sector'] || 'Technology').trim(),\n ticker: (form['Stock Ticker'] || '').trim(),\n github_token: (form['GitHub Token'] || '').trim(),\n groq_api_key: (form['Groq API Key'] || '').trim(),\n config,\n graphql_query,\n run_ts: new Date().toISOString(),\n run_ts_ms: Date.now(),\n }\n}];" }, "id": "cb5417fd-2d1e-4546-bd4f-5c669b8c6d5e", "name": "Company Input", "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ -272, -96 ] }, { "parameters": { "method": "POST", "url": "https://api.github.com/graphql", "authentication": "predefinedCredentialType", "nodeCredentialType": "githubApi", "sendBody": true, "contentType": "raw", "rawContentType": "application/json", "body": "={{ JSON.stringify({ query: $json.graphql_query }) }}", "options": { "response": { "response": { "neverError": true } } } }, "id": "04abae46-3c72-4222-b948-508ac05c6c53", "name": "GitHub GraphQL", "type": "n8n-nodes-base.httpRequest", "typeVersion": 4.2, "position": [ 32, -96 ] }, { "parameters": { "jsCode": "const item = $input.first().json; // GraphQL response\nconst ctx = $('Company Input').first().json; // original context\nconst cfg = ctx.config;\n\n// ── Extract repos from GraphQL response ───────────────────────────\nconst gql = item.data && item.data.organization;\nif (!gql || !gql.repositories) {\n return [{ json: {\n ...item,\n pipeline_status: 'GQL_ERROR',\n error: JSON.stringify(item.errors || item).slice(0, 400),\n }}];\n}\n\nconst raw = gql.repositories.nodes || [];\nconst now = Date.now();\nconst DAY = 86400000;\nconst recCut = now - cfg.recent_months * 30 * DAY;\nconst baseCut = now - cfg.lookback_months * 30 * DAY;\nconst basePeriod = cfg.lookback_months - cfg.recent_months;\n\n// ── Normalise every repo ──────────────────────────────────────────\nconst repos = raw.map(r => {\n const topics = (r.repositoryTopics?.nodes || []).map(n => n.topic?.name || '');\n const langEdges = r.languages?.edges || [];\n const totalBytes = r.languages?.totalSize || 1;\n const langDist = {};\n for (const e of langEdges) {\n langDist[e.node.name] = Math.round(e.size / totalBytes * 1000) / 10; // % of bytes\n }\n return {\n name: r.name,\n full_name: r.nameWithOwner,\n stars: r.stargazerCount || 0,\n forks: r.forkCount || 0,\n is_fork: r.isFork,\n created_ms: r.createdAt ? new Date(r.createdAt).getTime() : 0,\n pushed_ms: r.pushedAt ? new Date(r.pushedAt ).getTime() : 0,\n primary_lang: r.primaryLanguage?.name || null,\n lang_dist: langDist,\n topics,\n days_since_push: r.pushedAt ? Math.round((now - new Date(r.pushedAt).getTime()) / DAY) : 9999,\n };\n});\n\nconst nonFork = repos.filter(r => !r.is_fork);\nconst totalStars = nonFork.reduce((s, r) => s + r.stars, 0);\nconst active90 = nonFork.filter(r => r.days_since_push <= 90 ).length;\nconst stale365 = nonFork.filter(r => r.days_since_push > 365).length;\n\n// ── Time cohorts ──────────────────────────────────────────────────\nconst recNF = nonFork.filter(r => r.created_ms >= recCut);\nconst oldNF = nonFork.filter(r => r.created_ms >= baseCut && r.created_ms < recCut);\n\nconst mRate = recNF.length / cfg.recent_months;\nconst mBase = oldNF.length / basePeriod || 0;\nconst bRatio = mBase > 0 ? mRate / mBase : 0;\nconst hasBurst = bRatio >= cfg.burst_multiplier;\n\n// ── Star-weighted language distribution per cohort ────────────────\nfunction weightedLangDist(cohort) {\n const agg = {};\n let totalW = 0;\n for (const r of cohort) {\n if (r.stars < cfg.min_stars) continue;\n const w = Math.max(1, r.stars);\n totalW += w;\n for (const [lang, pct] of Object.entries(r.lang_dist)) {\n agg[lang] = (agg[lang] || 0) + pct * w;\n }\n }\n if (!totalW) return {};\n return Object.fromEntries(\n Object.entries(agg).map(([k, v]) => [k, Math.round(v / totalW * 10) / 10])\n );\n}\n\nconst recDist = weightedLangDist(recNF);\nconst oldDist = weightedLangDist(oldNF);\nconst allLangs = new Set([...Object.keys(recDist), ...Object.keys(oldDist)]);\n\n// ── Language deltas ───────────────────────────────────────────────\nconst deltas = [];\nfor (const lang of allLangs) {\n const rec = recDist[lang] || 0;\n const old = oldDist[lang] || 0;\n const diff = rec - old;\n const rel = old > 0 ? Math.abs(diff) / old * 100 : (rec > 0.5 ? 100 : 0);\n const isNew = old < 0.1 && rec > 0.5;\n const isDropped = rec < 0.1 && old > 0.5;\n if (rel >= cfg.lang_delta_pct || isNew || isDropped) {\n deltas.push({\n language: lang,\n recent_pct: Math.round(rec * 10) / 10,\n older_pct: Math.round(old * 10) / 10,\n delta_pp: Math.round(diff * 10) / 10,\n rel_chg: Math.round(rel),\n direction: isNew ? 'EMERGING' : isDropped ? 'ABANDONED' : diff > 0 ? 'RISING' : 'DECLINING',\n is_new: isNew,\n is_dropped: isDropped,\n });\n }\n}\ndeltas.sort((a, b) => Math.abs(b.delta_pp) - Math.abs(a.delta_pp));\n\n// ── AI/ML topic focus ratio ───────────────────────────────────────\nconst AI_TAGS = new Set(['machine-learning','deep-learning','llm','large-language-model',\n 'neural-network','nlp','computer-vision','reinforcement-learning','generative-ai',\n 'transformer','diffusion','multimodal','ai','pytorch','tensorflow','cuda','inference','fine-tuning']);\nconst aiRatio = (cohort) =>\n cohort.filter(r => r.topics.some(t => AI_TAGS.has(t))).length / Math.max(1, cohort.length);\nconst aiRec = aiRatio(recNF);\nconst aiOld = aiRatio(oldNF);\n\n// ── Top repos for velocity analysis ──────────────────────────────\nconst topRepos = nonFork\n .filter(r => r.stars >= cfg.min_stars)\n .sort((a, b) => b.stars - a.stars)\n .slice(0, cfg.velocity_repos)\n .map(r => ({\n name: r.name,\n full_name: r.full_name,\n stars: r.stars,\n primary_lang: r.primary_lang,\n days_since_push:r.days_since_push,\n commit_url: `https://api.github.com/repos/${r.full_name}/stats/commit_activity`,\n }));\n\nreturn [{\n json: {\n // carry forward\n name: item.name,\n org: item.org,\n sector: item.sector,\n ticker: item.ticker,\n config: item.config,\n run_ts: item.run_ts,\n pipeline_status: 'OK',\n\n repo_stats: {\n total_fetched: raw.length,\n non_fork: nonFork.length,\n total_stars: totalStars,\n avg_stars: nonFork.length ? Math.round(totalStars / nonFork.length) : 0,\n active_90d: active90,\n stale_365d: stale365,\n rec_cohort_size: recNF.length,\n old_cohort_size: oldNF.length,\n },\n burst: {\n has_burst: hasBurst,\n ratio: Math.round(bRatio * 100) / 100,\n monthly_rec: Math.round(mRate * 10) / 10,\n monthly_base: Math.round(mBase * 10) / 10,\n rec_count: recNF.length,\n signal: hasBurst\n ? `BURST DETECTED: ${recNF.length} new repos in ${cfg.recent_months}m vs ${Math.round(mBase*10)/10}/m historical baseline (${Math.round(bRatio*10)/10}x)`\n : 'NORMAL — repo creation rate within historical baseline',\n },\n language: {\n recent_dist: recDist,\n older_dist: oldDist,\n deltas,\n shift_detected: deltas.length > 0,\n rising: deltas.filter(d => d.direction === 'RISING' || d.direction === 'EMERGING'),\n declining: deltas.filter(d => d.direction === 'DECLINING' || d.direction === 'ABANDONED'),\n new_langs: deltas.filter(d => d.is_new),\n dropped: deltas.filter(d => d.is_dropped),\n },\n ai_focus: {\n recent_ratio: Math.round(aiRec * 100),\n older_ratio: Math.round(aiOld * 100),\n delta_pp: Math.round((aiRec - aiOld) * 100),\n increasing: (aiRec - aiOld) > 0.1,\n },\n top_repos: topRepos,\n }\n}];\n" }, "id": "7859312f-ecd0-4e49-8d97-fa8b03bcb4a9", "name": "Process Repos + Language Analysis", "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 336, -96 ] }, { "parameters": { "jsCode": "const d = $input.first().json;\nconst token = d.github_token;\n\nconst HEADERS = {\n 'Authorization': `Bearer ${token}`,\n 'Accept': 'application/vnd.github+json',\n 'X-GitHub-Api-Version': '2022-11-28',\n 'User-Agent': 'TSDSA-v2/1.0',\n};\n\nasync function fetchRepoCommits(repo) {\n const since = new Date(Date.now() - 365 * 86400000).toISOString();\n const url = `https://api.github.com/repos/${repo.full_name}/commits?per_page=100&since=${since}`;\n try {\n const resp = await $helpers.httpRequest({ method: 'GET', url, headers: HEADERS });\n const commits = Array.isArray(resp) ? resp : [];\n return {\n repo: repo.name,\n full_name: repo.full_name,\n stars: repo.stars,\n language: repo.primary_lang,\n commits,\n fetch_status: 'OK',\n };\n } catch (e) {\n const status = e.message.includes('404') ? 'EMPTY_REPO'\n : e.message.includes('409') ? 'NO_COMMITS'\n : 'FAILED';\n return {\n repo: repo.name,\n full_name: repo.full_name,\n stars: repo.stars,\n language: repo.primary_lang,\n commits: [],\n fetch_status: status,\n error: e.message.slice(0, 120),\n };\n }\n}\n\nconst settled = await Promise.allSettled(\n (d.top_repos || []).map(r => fetchRepoCommits(r))\n);\n\nconst commit_data = settled.map(s =>\n s.status === 'fulfilled' ? s.value\n : { repo: '?', commits: [], fetch_status: 'FAILED', error: String(s.reason) }\n);\n\nconst ok_count = commit_data.filter(r => r.fetch_status === 'OK').length;\nconst velocity_fetch_status =\n ok_count === 0 ? 'FAILED'\n : ok_count < commit_data.length ? 'PARTIAL'\n : 'OK';\n\nreturn [{ json: { ...d, commit_data, velocity_fetch_status } }];" }, "id": "42b80e57-1115-457a-b3e8-48c5fc1d27d0", "name": "Fetch Commit Activity", "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 608, -96 ] }, { "parameters": { "language": "python", "pythonCode": "import json, math, time, datetime\n\ninp = items[0]['json']\ncfg = inp.get('config', {})\nZ_THRESH = cfg.get('velocity_z_thresh', 2.0)\nNOW_MS = inp.get('run_ts_ms') or int(time.time() * 1000)\nDAY_MS = 86400000\n\ndef commits_to_weekly_buckets(commits):\n buckets = {}\n cutoff = NOW_MS - 365 * DAY_MS\n for c in (commits or []):\n try:\n if not isinstance(c, dict):\n continue\n commit_obj = c.get('commit') or {}\n author_obj = commit_obj.get('author') if isinstance(commit_obj, dict) else None\n ts_str = (author_obj.get('date', '') if isinstance(author_obj, dict) else '') or ''\n if not ts_str:\n continue\n dt = datetime.datetime.fromisoformat(ts_str.replace('Z', '+00:00'))\n ms = dt.timestamp() * 1000\n if ms < cutoff:\n continue\n wk = dt.strftime('%Y-W%W')\n buckets[wk] = buckets.get(wk, 0) + 1\n except Exception:\n continue\n if not buckets:\n return []\n sorted_keys = sorted(buckets.keys())\n return [buckets.get(w, 0) for w in sorted_keys]\n\ndef analyse_repo(repo):\n if not isinstance(repo, dict):\n return {\n 'repo': '?', 'language': None, 'stars': 0,\n 'fetch_status': 'UNKNOWN', 'commit_count': 0,\n 'status': 'UNAVAILABLE', 'signal': 'Invalid repo object',\n 'is_anomaly': False, 'direction': None, 'z_score': None,\n 'baseline_wkly': None, 'recent_wkly': None, 'std_dev': None,\n 'momentum_pct': None, 'pct_of_peak': None\n }\n commits = repo.get('commits') or []\n totals = commits_to_weekly_buckets(commits)\n base = {\n 'repo': repo.get('repo', '?'),\n 'language': repo.get('language'),\n 'stars': repo.get('stars', 0),\n 'fetch_status': repo.get('fetch_status', 'UNKNOWN'),\n 'commit_count': len(commits),\n }\n if len(totals) < 8:\n return {**base,\n 'status': 'UNAVAILABLE',\n 'signal': f\"Insufficient data ({len(commits)} commits sampled)\",\n 'is_anomaly': False,\n 'direction': None,\n 'z_score': None,\n 'baseline_wkly': None,\n 'recent_wkly': None,\n 'std_dev': None,\n 'momentum_pct': None,\n 'pct_of_peak': None}\n recent_w = totals[-4:]\n baseline_w = totals[:-4]\n b_mean = sum(baseline_w) / len(baseline_w)\n b_var = sum((x - b_mean)**2 for x in baseline_w) / len(baseline_w)\n b_std = math.sqrt(b_var)\n r_mean = sum(recent_w) / len(recent_w)\n z = (r_mean - b_mean) / b_std if b_std > 0 else 0.0\n is_anomaly = abs(z) >= Z_THRESH\n direction = ('ACCELERATION' if r_mean > b_mean else 'DECELERATION') if is_anomaly else None\n momentum = None\n if len(totals) >= 8:\n last2 = sum(totals[-2:]) / 2\n prev2 = sum(totals[-4:-2]) / 2\n momentum = round((last2 - prev2) / prev2 * 100) if prev2 > 0 else None\n windows = [sum(totals[i:i+4]) for i in range(len(totals) - 3)]\n peak_win = max(windows) if windows else 0\n recent_sum = sum(recent_w)\n pct_peak = round(recent_sum / peak_win * 100) if peak_win > 0 else 0\n lang = repo.get('language') or '?'\n return {**base,\n 'status': 'OK',\n 'is_anomaly': is_anomaly,\n 'direction': direction,\n 'z_score': round(z, 2),\n 'baseline_wkly': round(b_mean, 1),\n 'recent_wkly': round(r_mean, 1),\n 'std_dev': round(b_std, 1),\n 'momentum_pct': momentum,\n 'pct_of_peak': pct_peak,\n 'signal': (f\"VELOCITY {direction}: z={round(z,2)} in {repo.get('repo','?')} ({lang})\"\n if is_anomaly else\n f\"NORMAL — {repo.get('repo','?')} within baseline (z={round(z,2)})\")}\n\nsignals = [analyse_repo(r) for r in (inp.get('commit_data') or [])]\nanomalies = [s for s in signals if s.get('is_anomaly')]\naccel = [s for s in signals if s.get('direction') == 'ACCELERATION']\ndecel = [s for s in signals if s.get('direction') == 'DECELERATION']\nok_sigs = [s for s in signals if s.get('status') == 'OK']\n\nif not signals or len(ok_sigs) == 0:\n vel_status = 'UNAVAILABLE'\nelif len(ok_sigs) < len(signals):\n vel_status = 'PARTIAL'\nelse:\n vel_status = 'OK'\n\ntrend = ('ACCELERATING' if len(accel) > len(decel)\n else 'DECELERATING' if len(decel) > len(accel)\n else 'STABLE')\n\nreturn [{\n 'json': {\n **inp,\n 'velocity': {\n 'status': vel_status,\n 'signals': signals,\n 'anomaly_count': len(anomalies),\n 'acceleration_count':len(accel),\n 'deceleration_count':len(decel),\n 'top_anomaly': anomalies[0] if anomalies else None,\n 'overall_trend': trend,\n }\n }\n}]" }, "id": "7e06e112-953f-47e7-bdab-b5a46f4720a3", "name": "Velocity Analysis", "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 896, -96 ] }, { "parameters": { "jsCode": "const ctx = $('Company Input').first().json;\nconst d = { ...$input.first().json, config: ctx.config, name: ctx.name, org: ctx.org };\n\n// ── Compose data summary for the model ────────────────────────────\nconst topLangsStr = Object.entries(d.language.recent_dist || {})\n .sort(([,a],[,b]) => b - a).slice(0, 6)\n .map(([l, p]) => `${l}:${p}%`).join(', ') || 'n/a';\n\nconst deltaStr = (d.language.deltas || []).slice(0, 6)\n .map(x => `${x.language} ${x.direction} (${x.rel_chg}% relative, ${x.delta_pp}pp)`)\n .join('; ') || 'none detected';\n\nconst velStr = (d.velocity.signals || [])\n .filter(s => s.status === 'analysed')\n .map(s => `${s.repo}[${s.language||'?'},${s.stars}★]: z=${s.z_score}, ${s.baseline_wkly}→${s.recent_wkly} commits/wk, momentum=${s.momentum_pct !== null ? s.momentum_pct+'%' : 'n/a'}`)\n .join(' | ') || 'no velocity data';\n\nconst topReposStr = (d.top_repos || [])\n .map(r => `${r.name}(${r.stars}★,${r.primary_lang||'?'},${r.days_since_push}d)`)\n .join(', ');\n\nconst systemPrompt =\n 'You are a senior investment analyst specialising in AI/technology companies. ' +\n 'You analyse engineering signals from GitHub to identify technology strategy direction. ' +\n 'Respond ONLY with valid JSON — no markdown, no explanation, no code fences.';\n\nconst userPrompt =\n `Analyse the following GitHub engineering signals for ${d.name} (${d.sector}):\n\nCOMPANY: ${d.name} | ORG: ${d.org} | SECTOR: ${d.sector}\nREPO HEALTH: ${d.repo_stats.non_fork} public repos | ${d.repo_stats.total_stars.toLocaleString()} total stars | ${d.repo_stats.active_90d} active (90d) | ${d.repo_stats.stale_365d} stale (365d+)\nRECENT COHORT: ${d.repo_stats.rec_cohort_size} repos created in last ${d.config.recent_months} months\nOLDER COHORT: ${d.repo_stats.old_cohort_size} repos created in prior ${d.config.lookback_months - d.config.recent_months} months\n\nREPO CREATION BURST: ${d.burst.signal}\nAI/ML TOPIC FOCUS: recent=${d.ai_focus.recent_ratio}% repos | older=${d.ai_focus.older_ratio}% repos | delta=${d.ai_focus.delta_pp}pp\n\nTOP LANGUAGES (recent, star-weighted): ${topLangsStr}\nLANGUAGE SHIFTS (≥${d.config.lang_delta_pct}% relative change): ${deltaStr}\n\nCOMMIT VELOCITY (top repos by stars):\n${velStr}\n\nTOP REPOS: ${topReposStr}\n\nReturn ONLY this JSON:\n{\n \"signal_summary\": \"3-4 sentence investment-grade assessment of engineering direction\",\n \"velocity_assessment\": \"1-2 sentences on commit momentum and what it implies\",\n \"technology_direction\": \"Where is engineering effort moving? What does the language/repo data suggest about strategy?\",\n \"key_risks\": [\"risk 1\", \"risk 2\", \"risk 3\"],\n \"key_opportunities\": [\"opportunity 1\", \"opportunity 2\", \"opportunity 3\"],\n \"comparable_moves\": \"Has any comparable company made a similar engineering shift? Brief 1-sentence context.\",\n \"flags\": [\"FLAG1\", \"FLAG2\"],\n \"intensity\": \"NONE or LOW or MEDIUM or HIGH or CRITICAL\",\n \"confidence_score\": 0-100,\n \"composite_signal_score\": 0-100,\n \"investment_implication\": \"BULLISH or BEARISH or NEUTRAL or WATCH\"\n}`;\n\nreturn [{\n json: {\n ...d,\n groq_payload: JSON.stringify({\n model: 'llama-3.1-8b-instant',\n messages: [\n { role: 'system', content: systemPrompt },\n { role: 'user', content: userPrompt },\n ],\n temperature: 0.2,\n max_tokens: 900,\n response_format: { type: 'json_object' },\n }),\n }\n}];\n" }, "id": "aba79be3-5e28-466f-9db0-58cba81c15a6", "name": "Build Groq Prompt", "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 1168, -96 ] }, { "parameters": { "method": "POST", "url": "https://api.groq.com/openai/v1/chat/completions", "authentication": "predefinedCredentialType", "nodeCredentialType": "groqApi", "sendBody": true, "contentType": "raw", "rawContentType": "application/json", "body": "={{ $json.groq_payload }}", "options": { "response": { "response": { "neverError": true } } } }, "id": "b87cfba3-0f7f-42ec-b6fa-d75c2cd741f9", "name": "Groq AI Analysis", "type": "n8n-nodes-base.httpRequest", "typeVersion": 4.2, "position": [ 1456, -96 ] }, { "parameters": { "jsCode": "const groqResponse = $input.first().json; // Groq output — choices only\nconst cleanPipe = $('Build Groq Prompt').first().json; // full pipeline data\n\n// ── Safe deep getter ──────────────────────────────────────────────────────────\nconst g = (obj, ...path) => {\n let cur = obj;\n for (const k of path) {\n if (cur == null) return undefined;\n cur = Array.isArray(cur) && typeof k === 'number' ? cur[k] : (cur[k] ?? undefined);\n }\n return cur;\n};\nconst gd = (obj, ...args) => {\n const def = args[args.length - 1];\n const path = args.slice(0, -1);\n const val = g(obj, ...path);\n return val != null ? val : def;\n};\n\nconst cfg = cleanPipe.config || {};\nconst BURST_THRESHOLD = cfg.burst_multiplier ?? 2.0;\nconst LANG_DELTA_PCT = cfg.lang_delta_pct ?? 15;\nconst VEL_Z_THRESH = cfg.velocity_z_thresh ?? 2.0;\n\n// ── Parse Groq response ───────────────────────────────────────────────────────\nlet ai = {};\ntry {\n const choices = groqResponse.choices || [];\n const content = choices[0]?.message?.content || '{}';\n ai = JSON.parse(content);\n} catch(e) { ai = {}; }\n\nconst AI_DEFAULTS = {\n signal_summary: 'AI analysis unavailable — check Groq API key',\n velocity_assessment: 'Velocity data could not be retrieved.',\n technology_direction: 'Insufficient data for direction signal.',\n key_risks: ['Data quality risk: velocity endpoint unavailable'],\n key_opportunities: [],\n comparable_moves: '',\n flags: [],\n intensity: 'NONE',\n confidence_score: 0,\n composite_signal_score:0,\n investment_implication:'NEUTRAL',\n};\nfor (const [k, v] of Object.entries(AI_DEFAULTS)) {\n if (ai[k] == null) ai[k] = v;\n}\n\n// ── Rule flags ────────────────────────────────────────────────────────────────\nconst burstRatio = gd(cleanPipe, 'burst', 'ratio', 0);\nconst burstBase = gd(cleanPipe, 'burst', 'monthly_base', 0);\nconst hasBurst = burstBase > 0 && burstRatio >= BURST_THRESHOLD;\n\nconst ruleFlags = [];\nif (hasBurst) ruleFlags.push('REPO_BURST');\nif (gd(cleanPipe,'language','shift_detected',false)) ruleFlags.push('LANGUAGE_SHIFT');\nif (gd(cleanPipe,'ai_focus','increasing',false)) ruleFlags.push('AI_FOCUS_INCREASING');\nif (gd(cleanPipe,'velocity','acceleration_count',0) > 0) ruleFlags.push('VELOCITY_ACCELERATION');\nif (gd(cleanPipe,'velocity','deceleration_count',0) > 0) ruleFlags.push('VELOCITY_DECELERATION');\nif ((gd(cleanPipe,'language','new_langs',[]) || []).length > 0) ruleFlags.push('NEW_LANGUAGE_ADOPTION');\n\nconst allFlags = [...new Set([...ruleFlags, ...(ai.flags || [])])];\n\n// ── Composite score ───────────────────────────────────────────────────────────\nlet ruleScore = 0;\nif (hasBurst) ruleScore += Math.min(30, burstRatio * 10);\nconst deltas = gd(cleanPipe, 'language', 'deltas', []) || [];\nif (deltas.length) ruleScore += Math.min(25, deltas.length * 5);\nif (gd(cleanPipe,'ai_focus','increasing',false)) ruleScore += 10;\nconst anomalyCount = gd(cleanPipe,'velocity','anomaly_count',0);\nif (anomalyCount > 0) ruleScore += Math.min(20, anomalyCount * 10);\n\nconst composite = Math.min(100, Math.round(ai.composite_signal_score * 0.6 + ruleScore * 0.4));\n\n// ── Intensity ─────────────────────────────────────────────────────────────────\nconst ORD = ['NONE','LOW','MEDIUM','HIGH','CRITICAL'];\nconst ruleI = composite >= 85 ? 'CRITICAL' : composite >= 65 ? 'HIGH' :\n composite >= 45 ? 'MEDIUM' : composite >= 25 ? 'LOW' : 'NONE';\nconst groqI = ORD.includes(ai.intensity) ? ai.intensity : 'NONE';\nconst finalI = ORD.indexOf(groqI) >= ORD.indexOf(ruleI) ? groqI : ruleI;\n\nconst IC = {CRITICAL:'#dc2626',HIGH:'#ef4444',MEDIUM:'#f59e0b',LOW:'#3b82f6',NONE:'#4b5563'};\n\n// ── Evidence trail ────────────────────────────────────────────────────────────\nconst langEvidence = {\n category: 'language_shift',\n status: deltas.length ? 'DETECTED' : 'NONE',\n threshold_pct: LANG_DELTA_PCT,\n items: deltas.slice(0,5).map(d => ({\n language: d.language, from_pct: d.older_pct,\n to_pct: d.recent_pct, rel_change_pct: d.rel_chg, direction: d.direction,\n })),\n summary: deltas.length\n ? `${deltas.length} language(s) shifted >${LANG_DELTA_PCT}% relative`\n : `No language shifts detected at >${LANG_DELTA_PCT}% threshold`,\n};\n\nconst rs = cleanPipe.repo_stats || {};\nconst burst = cleanPipe.burst || {};\nconst repoEvidence = {\n category: 'repo_growth',\n status: hasBurst ? 'BURST' : 'NORMAL',\n burst_ratio: Math.round(burstRatio * 100) / 100,\n burst_threshold: BURST_THRESHOLD,\n monthly_recent: burst.monthly_rec ?? 0,\n monthly_baseline: burst.monthly_base ?? 0,\n total_repos: rs.non_fork || 0,\n active_90d: rs.active_90d || 0,\n stale_365d: rs.stale_365d || 0,\n summary: burst.signal || 'Repo growth data unavailable',\n};\n\nconst vel = cleanPipe.velocity || {};\nconst velStatus = vel.status || 'UNAVAILABLE';\nconst velSigs = (vel.signals || []).filter(s => s.status === 'OK');\nconst velEvidence = {\n category: 'velocity_signal',\n status: velStatus,\n overall_trend: vel.overall_trend || 'UNKNOWN',\n repos_analysed: velSigs.length,\n repos_total: (vel.signals || []).length,\n anomalies_detected: vel.anomaly_count || 0,\n top_anomaly: vel.top_anomaly || null,\n items: velSigs.slice(0,3).map(s => ({\n repo: s.repo, z_score: s.z_score,\n baseline_wkly: s.baseline_wkly, recent_wkly: s.recent_wkly,\n direction: s.direction, momentum_pct: s.momentum_pct,\n })),\n summary: velStatus !== 'UNAVAILABLE'\n ? `Velocity ${velStatus}: ${vel.overall_trend || '?'}, ${vel.anomaly_count || 0} anomalies across ${velSigs.length} repos analysed`\n : 'Commit velocity data unavailable — endpoint did not return sufficient data',\n};\n\nconst anomalyEvidence = {\n category: 'anomaly_detection',\n status: (vel.anomaly_count || 0) > 0 ? 'DETECTED' : 'NONE',\n acceleration_count: vel.acceleration_count || 0,\n deceleration_count: vel.deceleration_count || 0,\n items: (vel.signals || []).filter(s => s.is_anomaly).map(s => ({\n repo: s.repo, direction: s.direction, z_score: s.z_score, signal: s.signal,\n })),\n summary: (vel.anomaly_count || 0) > 0\n ? `${vel.anomaly_count} velocity anomaly(ies) detected`\n : 'No velocity anomalies at current z-score threshold',\n};\n\n// ── Claims schema v2.0 ────────────────────────────────────────────────────────\nconst name = cleanPipe.name || 'UNKNOWN';\nconst org = cleanPipe.org || '?';\nconst ts = cleanPipe.run_ts || '';\n\nconst claim = {\n claim_id: `TSDSA-${name.replace(/\\s+/g,'_').toUpperCase()}-${ts.slice(0,10)}`,\n agent: 'TechStackDirectionalSignalAgent',\n schema_version: '2.0',\n timestamp: ts,\n subject: name,\n github_org: org,\n sector: cleanPipe.sector || '',\n claim_type: 'ENGINEERING_DIRECTION_DELTA',\n intensity: finalI,\n intensity_color:IC[finalI] || '#4b5563',\n composite_score:composite,\n confidence: ai.confidence_score,\n confidence_degraded: velStatus === 'UNAVAILABLE',\n investment_implication: ai.investment_implication,\n flags: allFlags,\n signals: {\n velocity_status: velStatus,\n language_shift: deltas.length > 0,\n burst_detected: hasBurst,\n ai_focus: gd(cleanPipe,'ai_focus','increasing',false),\n },\n ai_analysis: {\n signal_summary: ai.signal_summary,\n velocity_assessment: ai.velocity_assessment,\n technology_direction: ai.technology_direction,\n comparable_moves: ai.comparable_moves,\n key_risks: ai.key_risks,\n key_opportunities: ai.key_opportunities,\n },\n engineering_direction_delta: {\n rising_languages: (gd(cleanPipe,'language','rising', []) || []).map(x => x.language),\n declining_languages: (gd(cleanPipe,'language','declining',[]) || []).map(x => x.language),\n new_languages: (gd(cleanPipe,'language','new_langs',[]) || []).map(x => x.language),\n abandoned_languages: (gd(cleanPipe,'language','dropped', []) || []).map(x => x.language),\n repo_trend: hasBurst ? 'ACCELERATING' : 'NORMAL',\n commit_trend: vel.overall_trend || 'STABLE',\n velocity_status: velStatus,\n },\n evidence: {\n language_shift: langEvidence,\n repo_growth: repoEvidence,\n velocity_signal: velEvidence,\n anomaly_detection: anomalyEvidence,\n },\n risk_flags: ai.key_risks,\n opportunity_flags: ai.key_opportunities,\n resolver_ready: true,\n human_review: ['HIGH','CRITICAL'].includes(finalI),\n};\n\nreturn [{ json: { ...cleanPipe, claim } }];" }, "id": "8443750b-e60a-417e-8bbc-ffa07aaed0fe", "name": "Score + Claims Schema", "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 1728, -96 ] }, { "parameters": { "jsCode": "const velData = $('Velocity Analysis').first().json;\nconst d = { ...velData, ...$input.first().json };\nconst cl = d.claim;\nconst ai = cl.ai_analysis;\nconst edd = cl.engineering_direction_delta;\nconst la = d.language;\nconst bd = d.burst;\nconst va = d.velocity;\nconst rs = d.repo_stats;\nconst genAt = new Date(d.run_ts).toLocaleString('en-US',{dateStyle:'medium',timeStyle:'short'});\n\nconst IC = {CRITICAL:'#dc2626',HIGH:'#ef4444',MEDIUM:'#f59e0b',LOW:'#3b82f6',NONE:'#4b5563'};\nconst IB = {CRITICAL:'#2d0a0a',HIGH:'#1f0a0a',MEDIUM:'#1f1500',LOW:'#0a1020',NONE:'#111827'};\nconst FC = {\n REPO_BURST:'#8b5cf6', LANGUAGE_SHIFT:'#06b6d4', AI_FOCUS_INCREASING:'#10b981',\n VELOCITY_ACCELERATION:'#22c55e', VELOCITY_DECELERATION:'#f97316',\n NEW_LANGUAGE_ADOPTION:'#f59e0b',\n};\nconst IMPL_C = {BULLISH:'#10b981',BEARISH:'#ef4444',WATCH:'#f59e0b',NEUTRAL:'#6b7280'};\n\nconst ic = k => IC[k] || '#4b5563';\nconst ibc = k => IB[k] || '#111827';\nconst fc = k => FC[k] || '#6b7280';\nconst col = ic(cl.intensity);\nconst bg = ibc(cl.intensity);\n\nconst pill = (text, c) =>\n `${text.replace(/_/g,' ')}`;\n\nconst implColor = IMPL_C[cl.investment_implication] || '#6b7280';\nconst implPill = `` +\n `${cl.investment_implication}`;\n\n// ── Language delta rows ────────────────────────────────────────────\nconst langRows = (la.deltas || []).slice(0, 8).map(d => {\n const c = d.direction === 'RISING' || d.direction === 'EMERGING' ? '#10b981' : '#ef4444';\n const icon = d.is_new ? '✦ NEW' : d.is_dropped ? '✕ GONE' :\n d.direction === 'RISING' ? '▲' : '▼';\n const c2 = d.is_new ? '#8b5cf6' : d.is_dropped ? '#4b5563' : c;\n return `\n ${d.language}\n ${d.older_pct}%\n ${d.recent_pct}%\n \n ${icon} ${d.rel_chg}% rel.\n `;\n}).join('') || `\n No shifts detected at ≥${d.config?.lang_delta_pct || 15}% relative threshold`;\n\n// ── Velocity rows ──────────────────────────────────────────────────\nconst velRows = (va.signals || []).filter(s => s.status === 'analysed').map(s => {\n const c = s.direction === 'ACCELERATION' ? '#10b981' : s.direction === 'DECELERATION' ? '#ef4444' : '#4b5563';\n const icon = s.is_anomaly ? (s.direction === 'ACCELERATION' ? '▲' : '▼') : '○';\n const mPill = s.momentum_pct !== null\n ? `0?'#10b981':'#ef4444'};font-size:9px\">${s.momentum_pct>0?'+':''}${s.momentum_pct}% momentum`\n : '';\n return `
\n
\n ${s.repo}\n ${icon} ${s.direction || 'NORMAL'}\n
\n
\n ${s.language || '?'} · ${(s.stars||0).toLocaleString()}★\n z = ${s.z_score}\n ${s.baseline_wkly} → ${s.recent_wkly} commits/wk\n ${mPill}\n ${s.pct_of_peak}% of peak\n
\n
`;\n}).join('') || '
No commit data available
';\n\n// ── Evidence pills ────────────────────────────────────────────────\nconst EV_STATUS_COLOR = {DETECTED:'#10b981',BURST:'#8b5cf6',OK:'#10b981',PARTIAL:'#f59e0b',NONE:'#4b5563',UNAVAILABLE:'#6b7280',NORMAL:'#3b82f6'};\nconst evidenceHtml = ['language_shift','repo_growth','velocity_signal','anomaly_detection'].map(key => {\n const e = (cl.evidence || {})[key] || {};\n const sc = EV_STATUS_COLOR[e.status] || '#4b5563';\n return `
\n
${key.replace(/_/g,' ')} · ${e.status || 'NO DATA'}
\n
${e.summary || 'No summary available'}
\n
`;\n}).join('');\n\n// ── Top repos ─────────────────────────────────────────────────────\nconst topRepoRows = (d.top_repos || []).slice(0, 6).map(r =>\n `
\n ${r.name}\n
\n ${(r.stars||0).toLocaleString()}★\n ${r.primary_lang || '?'}\n ${r.days_since_push}d ago\n
\n
`\n).join('');\n\nconst flagsHtml = (cl.flags || []).length\n ? (cl.flags || []).map(f => pill(f, fc(f))).join(' ')\n : `no flags raised`;\n\nconst riskHtml = (ai.key_risks || []).map(r =>\n `
⚠ ${r}
`\n).join('') || '
No risks flagged by AI
';\n\nconst oppHtml = (ai.key_opportunities || []).map(o =>\n `
✓ ${o}
`\n).join('') || '
No opportunities flagged
';\n\nconst dirParts = [];\nif (edd.what_increased?.length) dirParts.push(`▲ RISING: ${edd.what_increased.join(', ')}`);\nif (edd.what_decreased?.length) dirParts.push(`▼ DECLINING: ${edd.what_decreased.join(', ')}`);\nif (edd.newly_appeared?.length) dirParts.push(`✦ NEW: ${edd.newly_appeared.join(', ')}`);\nif (edd.stalled_or_dropped?.length)dirParts.push(`✕ DROPPED: ${edd.stalled_or_dropped.join(', ')}`);\nconst dirHtml = dirParts.join('
') || 'No directional shift detected';\n\nconst html = `\n\n\n\n\n${d.name} — Engineering Direction Delta · Mycroft\n\n\n\n\n\n
\n\n \n
\n
\n
\n Mycroft Framework · Intelligence Layer · Tech Stack Directional Signal Agent\n
\n
\n Engineering Direction Delta\n
\n
\n ${d.name}${d.ticker ? ' (' + d.ticker + ')' : ''}\n
\n
\n github.com/${d.org} · ${d.sector} · Run: ${genAt}\n
\n
\n
\n
${cl.composite_score}
\n
COMPOSITE SIGNAL SCORE
\n
${cl.intensity}
\n
${implPill}
\n
\n
\n\n \n
\n
\n
\n\n \n
\n SIGNAL FLAGS\n ${flagsHtml}\n \n Confidence: ${cl.confidence}% ·\n Human review: ${cl.human_review?'REQUIRED':'not required'} ·\n Resolver: READY\n \n
\n\n \n
\n
\n AI SIGNAL ASSESSMENT · Groq llama-3.1-8b-instant\n
\n
\n ${ai.signal_summary || '—'}\n
\n
\n
\n
TECHNOLOGY DIRECTION
\n
${ai.technology_direction || '—'}
\n
\n
\n
VELOCITY ASSESSMENT
\n
${ai.velocity_assessment || '—'}
\n
\n
\n ${ai.comparable_moves ? `\n
\n
COMPARABLE MARKET MOVES
\n
${ai.comparable_moves}
\n
` : ''}\n
\n\n \n
\n\n \n
\n
\n LANGUAGE SHIFTS (≥${d.config?.lang_delta_pct || 15}% REL.)\n
\n \n \n \n \n \n \n \n ${langRows}\n
LANGOLDERRECENTSHIFT
\n
\n
DIRECTION SUMMARY
\n
${dirHtml}
\n
\n
\n\n \n
\n
\n COMMIT VELOCITY (z ≥ ${d.config?.velocity_z_thresh || 2}σ)\n
\n ${velRows}\n
\n
OVERALL TREND
\n
\n ${va.overall_trend || 'STABLE'}\n
\n
\n ${va.acceleration_count} accelerating · ${va.deceleration_count} decelerating · ${va.anomaly_count} anomalies\n
\n
\n
\n\n \n
\n
KEY RISKS
\n ${riskHtml}\n
\n
OPPORTUNITIES
\n ${oppHtml}\n
\n
\n\n \n
\n
REPO HEALTH
\n ${[\n ['Non-fork repos', rs.non_fork, '#e2e8f0'],\n ['Total stars', (rs.total_stars||0).toLocaleString(), '#f59e0b'],\n ['Active (90d)', rs.active_90d, '#10b981'],\n ['Stale (365d+)', rs.stale_365d, '#ef4444'],\n ['Recent cohort', rs.rec_cohort_size + ' repos / ' + d.config?.recent_months + 'm', '#94a3b8'],\n ['Older cohort', rs.old_cohort_size + ' repos / ' + (d.config?.lookback_months - d.config?.recent_months) + 'm', '#94a3b8'],\n ['Burst ratio', bd.ratio + 'x baseline', bd.has_burst ? '#f59e0b' : '#4b5563'],\n ['AI topic focus', d.ai_focus?.recent_ratio + '% recent vs ' + d.ai_focus?.older_ratio + '% older', '#06b6d4'],\n ].map(([label, val, vc]) =>\n `
\n ${label}\n ${val}\n
`\n ).join('')}\n\n
\n
TOP REPOS
\n ${topRepoRows}\n
\n
\n\n
\n\n \n
\n
\n EVIDENCE TRAIL — Contradiction Resolver Input\n
\n
\n ${evidenceHtml}\n
\n
\n\n \n
\n
\n MYCROFT CLAIMS SCHEMA OUTPUT — claim_id: ${cl.claim_id}\n
\n 
${JSON.stringify({\n      claim_id:        cl.claim_id,\n      agent:           cl.agent,\n      schema_version:  cl.schema_version,\n      subject:         cl.subject,\n      claim_type:      cl.claim_type,\n      intensity:       cl.intensity,\n      composite_score: cl.composite_score,\n      investment_implication: cl.investment_implication,\n      flags:           cl.flags,\n      engineering_direction_delta: cl.engineering_direction_delta,\n      resolver_ready:  cl.resolver_ready,\n      human_review:    cl.human_review,\n    }, null, 2)}
\n
\n\n \n
\n Mycroft Framework · Tech Stack Directional Signal Agent v3 · ${genAt}\n Schema v1.0 · GitHub GraphQL + Groq llama-3.1-8b-instant · Resolver-Ready\n
\n\n
\n\n`;\n\nreturn [{ json: { html } }];\n" }, "id": "724fa27a-5188-4711-ad23-5d1fbc4a173f", "name": "Generate Dashboard", "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 2016, -96 ] }, { "parameters": { "jsCode": "const html = $input.first().json.html;\nconst buf = Buffer.from(html, 'utf8');\nconst bin = await this.helpers.prepareBinaryData(buf, 'engineering_direction_delta.html', 'text/html');\nreturn [{\n json: { filename: 'engineering_direction_delta.html', size_kb: Math.round(buf.length / 1024) },\n binary: { data: bin },\n}];" }, "id": "e32978f2-aa4c-4ca6-995c-15d65fa34411", "name": "Export HTML File", "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 2288, -96 ] }, { "parameters": { "jsCode": "return [{\n json: {\n message: '✅ Report generated! Go to n8n → Export HTML File → Output → Binary → Download'\n }\n}];" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 2496, -96 ], "id": "1c3908a7-acf0-47b6-8007-5fe397ed7550", "name": "Code" }, { "id": "973f7dba-d5de-5c09-a34f-1fa5e33e2434", "name": "Publication · Overview and setup", "type": "n8n-nodes-base.stickyNote", "typeVersion": 1, "position": [ -1328, -96 ], "parameters": { "content": "# Tech_Stack_Directional_Signal_Agent\n\n## Purpose and structure\nThis folder contains the **Tech_Stack_Directional_Signal_Agent** workflow. Its executable nodes use `code`, `formTrigger`, `httpRequest`. The table and connection map below are derived from the JSON, not an inferred guarantee of behavior.\n\n## Setup\nImport inactive. Configure your own credentials and replace REPLACE_WITH_* values. Review external destinations and Code nodes before running.\n\n## Entry points\nForm Trigger\n\n## Verification\nSanitized source candidate; not runtime-tested. This overview is generated from node metadata. Read the folder README for the connection map, credential types, and license.\n\n## Attribution\nNik Bear Brown; rights confirmed 2026-10-19\nMIT; preserve source notices.", "width": 700, "height": 800, "color": 1 } } ], "connections": { "Company Input": { "main": [ [ { "node": "GitHub GraphQL", "type": "main", "index": 0 } ] ] }, "GitHub GraphQL": { "main": [ [ { "node": "Process Repos + Language Analysis", "type": "main", "index": 0 } ] ] }, "Process Repos + Language Analysis": { "main": [ [ { "node": "Fetch Commit Activity", "type": "main", "index": 0 } ] ] }, "Fetch Commit Activity": { "main": [ [ { "node": "Velocity Analysis", "type": "main", "index": 0 } ] ] }, "Velocity Analysis": { "main": [ [ { "node": "Build Groq Prompt", "type": "main", "index": 0 } ] ] }, "Build Groq Prompt": { "main": [ [ { "node": "Groq AI Analysis", "type": "main", "index": 0 } ] ] }, "Groq AI Analysis": { "main": [ [ { "node": "Score + Claims Schema", "type": "main", "index": 0 } ] ] }, "Score + Claims Schema": { "main": [ [ { "node": "Generate Dashboard", "type": "main", "index": 0 } ] ] }, "Generate Dashboard": { "main": [ [ { "node": "Export HTML File", "type": "main", "index": 0 } ] ] }, "Form Trigger": { "main": [ [ { "node": "Company Input", "type": "main", "index": 0 } ] ] }, "Export HTML File": { "main": [ [ { "node": "Code", "type": "main", "index": 0 } ] ] } }, "active": false, "settings": { "executionOrder": "v1" } }