Screen Cvs with Openai and Postgresql Using Chained Prompts

The workflow JSON

Copy or download the full n8n JSON below. Paste it into a new n8n workflow, add your credentials, activate. Full import guide →

{
  "id": "FoMizxWit6hOJVzp",
  "meta": {
    "templateCredsSetupCompleted": true
  },
  "name": "AI CV Screening with Chained Prompts",
  "tags": [],
  "nodes": [
    {
      "id": "0cb0572a-b6e0-49e4-a9f2-c7249f9df88a",
      "name": "Overview",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -272,
        64
      ],
      "parameters": {
        "color": 3,
        "width": 500,
        "height": 832,
        "content": "## \ud83e\udd16 AI CV Screening with Chained Prompts\n\nAutomatically screen resumes using 4 sequential AI prompts, each building on the previous one's output. Results are saved directly to PostgreSQL \u2014 no external backend required.\n\n### How it works\n1. **Webhook** receives a job ID + list of candidate IDs. Candidates and job must already exist in the database.\n2. **Prompt 0** extracts a structured job template (requirements, differentials, behavioral competencies and weights) from the job description \u2014 runs only if `gabarito` is null in the jobs table.\n3. **Prompts 1\u20133** run for each candidate in a loop:\n   - **Prompt 1** scores the candidate (0\u2013100) against the job template with calibration anchors to avoid score inflation, plus per-criteria scores\n   - **Prompt 2** uses the score as context to identify concrete strengths (with CV evidence) and critical vs secondary gaps\n   - **Prompt 3** uses the gaps as context to generate personalized interview questions for that specific candidate\n4. Results are saved directly to the `analyses` table and candidate status updated to `processed`\n5. **Prompt 4** runs automatically when all candidates are processed \u2014 generates an executive summary saved to `job_summaries`\n\n### Setup\n- Add your **OpenAI API credentials** to all AI nodes\n- Add your **PostgreSQL credentials** to all Postgres nodes\n- Create the required tables using the SQL schema in the sticky note below\n- Trigger via `POST /webhook/cv-analyze` with `{ \"job_id\": 1, \"candidate_ids\": [1, 2, 3] }`\n\n### Customization\n- Swap `gpt-4.1-mini` for a more powerful model for higher accuracy\n- Adjust scoring anchors in Prompt 1 to match your hiring standards\n- Add more criteria to the job template in Prompt 0"
      },
      "typeVersion": 1
    },
    {
      "id": "5ef87983-e5bb-49f1-9ea7-7842acb476d2",
      "name": "Database Schema",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        800,
        672
      ],
      "parameters": {
        "width": 708,
        "height": 880,
        "content": "## \ud83d\uddc4\ufe0f Required Database Schema\n\n```sql\nCREATE TABLE jobs (\n  id SERIAL PRIMARY KEY,\n  title VARCHAR(255) NOT NULL,\n  description TEXT NOT NULL,\n  gabarito JSONB,\n  status VARCHAR(20) DEFAULT 'draft',\n  created_at TIMESTAMP DEFAULT NOW()\n);\n\nCREATE TABLE candidates (\n  id SERIAL PRIMARY KEY,\n  job_id INTEGER REFERENCES jobs(id) ON DELETE CASCADE,\n  name VARCHAR(255),\n  filename VARCHAR(255),\n  cv_text TEXT,\n  status VARCHAR(20) DEFAULT 'pending',\n  created_at TIMESTAMP DEFAULT NOW()\n);\n\nCREATE TABLE analyses (\n  id SERIAL PRIMARY KEY,\n  candidate_id INTEGER REFERENCES candidates(id) ON DELETE CASCADE,\n  job_id INTEGER REFERENCES jobs(id) ON DELETE CASCADE,\n  score INTEGER,\n  nivel_aderencia VARCHAR(20),\n  justificativa_score TEXT,\n  pontos_fortes JSONB,\n  gaps_criticos JSONB,\n  gaps_secundarios JSONB,\n  perguntas_entrevista JSONB,\n  score_criterios JSONB,\n  created_at TIMESTAMP DEFAULT NOW(),\n  CONSTRAINT analyses_candidate_unique UNIQUE (candidate_id)\n);\n\nCREATE TABLE job_summaries (\n  id SERIAL PRIMARY KEY,\n  job_id INTEGER REFERENCES jobs(id) ON DELETE CASCADE,\n  total_analisados INTEGER,\n  recomendados JSONB,\n  destaque TEXT,\n  gap_comum TEXT,\n  resumo TEXT,\n  created_at TIMESTAMP DEFAULT NOW(),\n  CONSTRAINT job_summaries_job_id_key UNIQUE (job_id)\n);\n```"
      },
      "typeVersion": 1
    },
    {
      "id": "e04df0d8-215c-4c7c-9c76-e49cd768cbdf",
      "name": "Input Format",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        304,
        64
      ],
      "parameters": {
        "width": 340,
        "height": 232,
        "content": "## \ud83d\udce5 Webhook Input\nSend a POST request:\n```json\n{\n  \"job_id\": 1,\n  \"candidate_ids\": [1, 2, 3]\n}\n```\nCandidates and job must already exist in the database with `cv_text` populated."
      },
      "typeVersion": 1
    },
    {
      "id": "e3297343-eac2-4658-a1a7-19d66f2df74f",
      "name": "Prompt 0 Section",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        832,
        176
      ],
      "parameters": {
        "width": 700,
        "height": 140,
        "content": "## \ud83e\udde0 Prompt 0 \u2014 Job Template Extraction\nRuns only when `gabarito` is null in the jobs table.\nExtracts structured requirements and sets weights automatically."
      },
      "typeVersion": 1
    },
    {
      "id": "68d2ff58-4421-422b-8a6b-008b40527bde",
      "name": "Candidate Loop Section",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        2112,
        64
      ],
      "parameters": {
        "width": 1700,
        "height": 140,
        "content": "## \ud83d\udd04 Candidate Loop\nProcesses one candidate at a time.\nPrompts 1 \u2192 2 \u2192 3 run sequentially,\neach using the previous output as context.\nResults saved to Postgres after each candidate."
      },
      "typeVersion": 1
    },
    {
      "id": "512b3240-cb2c-4d8a-9ad7-be87df671b62",
      "name": "Summary Section",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        4432,
        64
      ],
      "parameters": {
        "width": 1100,
        "height": 140,
        "content": "## \ud83d\udcca Executive Summary \u2014 Prompt 4\nRuns once when all candidates are processed (pending = 0).\nSaves pool-level recommendation to job_summaries table."
      },
      "typeVersion": 1
    },
    {
      "id": "7b48fa44-ea08-4c46-b89e-388d3ae9ee0e",
      "name": "Receive CVs",
      "type": "n8n-nodes-base.webhook",
      "position": [
        304,
        360
      ],
      "parameters": {
        "path": "cv-analyze",
        "options": {},
        "httpMethod": "POST"
      },
      "typeVersion": 2.1
    },
    {
      "id": "6dbd0cb2-dba2-46dd-a355-eb8bb474e68b",
      "name": "Fetch Job and Candidates",
      "type": "n8n-nodes-base.postgres",
      "position": [
        528,
        360
      ],
      "parameters": {
        "query": "SELECT j.id AS job_id, j.title, j.description, j.gabarito, array_agg(c.id) AS candidate_ids, json_agg(json_build_object('id', c.id, 'name', c.name, 'cv_text', c.cv_text)) AS candidates FROM jobs j JOIN candidates c ON c.job_id = j.id WHERE j.id = {{ $json.body.job_id }} AND c.id = ANY(ARRAY[{{ $json.body.candidate_ids.join(',') }}]::int[]) GROUP BY j.id, j.title, j.description, j.gabarito",
        "options": {},
        "operation": "executeQuery"
      },
      "credentials": {
        "postgres": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.5
    },
    {
      "id": "4cfed364-6856-4b72-a781-dbcc5e784879",
      "name": "Job Template exists?",
      "type": "n8n-nodes-base.if",
      "position": [
        752,
        360
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 2,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "id": "cond-gabarito",
              "operator": {
                "type": "object",
                "operation": "notEmpty",
                "singleValue": true
              },
              "leftValue": "={{ $json.gabarito }}",
              "rightValue": ""
            }
          ]
        }
      },
      "typeVersion": 2.2
    },
    {
      "id": "e0574622-0404-4739-8f86-64e9f2061a5a",
      "name": "Prompt 0 \u2014 Extract Job Template",
      "type": "@n8n/n8n-nodes-langchain.openAi",
      "position": [
        976,
        432
      ],
      "parameters": {
        "modelId": {
          "__rl": true,
          "mode": "list",
          "value": "gpt-4.1-mini",
          "cachedResultName": "GPT-4.1-MINI"
        },
        "options": {},
        "responses": {
          "values": [
            {
              "role": "system",
              "content": "You are an HR and recruitment specialist. Analyze the job description and extract information in a structured way. Return ONLY valid JSON, no additional text, no markdown, no backticks."
            },
            {
              "content": "=JOB DESCRIPTION:\n{{ $('Fetch Job and Candidates').item.json.description }}\n\nReturn ONLY valid JSON in this format:\n{\n  \"mandatory_requirements\": [\"list of mandatory requirements\"],\n  \"differential_requirements\": [\"list of differentials\"],\n  \"behavioral_competencies\": [\"list of behavioral competencies\"],\n  \"seniority_level\": \"junior|mid|senior|specialist\",\n  \"area\": \"area of expertise\",\n  \"weights\": {\n    \"mandatory_requirements\": 0.5,\n    \"differential_requirements\": 0.2,\n    \"behavioral_competencies\": 0.2,\n    \"experience\": 0.1\n  }\n}\n\nWeights must sum to 1.0 and reflect the importance of each criterion for THIS specific job."
            }
          ]
        },
        "builtInTools": {}
      },
      "credentials": {
        "openAiApi": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.1
    },
    {
      "id": "1f095e0e-bb1e-4eb7-b2bf-593e775928bb",
      "name": "Save Job Template",
      "type": "n8n-nodes-base.postgres",
      "position": [
        1328,
        432
      ],
      "parameters": {
        "query": "UPDATE jobs SET gabarito = '{{ $json.output[0].content[0].text }}'::jsonb WHERE id = {{ $('Fetch Job and Candidates').item.json.job_id }} RETURNING gabarito",
        "options": {},
        "operation": "executeQuery"
      },
      "credentials": {
        "postgres": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.5
    },
    {
      "id": "aed6cd32-e56a-482a-85ec-3a3ccde8e35f",
      "name": "Prepare Candidates",
      "type": "n8n-nodes-base.code",
      "position": [
        1552,
        360
      ],
      "parameters": {
        "jsCode": "const jobData = $('Fetch Job and Candidates').first().json;\n\nlet gabarito = jobData.gabarito;\n\nif (!gabarito) {\n  const raw = $('Prompt 0 \u2014 Extract Job Template').first().json.output[0].content[0].text;\n  gabarito = JSON.parse(raw);\n}\n\nconst candidates = typeof jobData.candidates === 'string'\n  ? JSON.parse(jobData.candidates)\n  : jobData.candidates;\n\nreturn candidates.map(c => ({\n  json: {\n    job_id:         jobData.job_id,\n    job_title:      jobData.title,\n    gabarito:       gabarito,\n    candidate_id:   c.id,\n    candidate_name: c.name,\n    cv_text:        c.cv_text\n  }\n}));"
      },
      "typeVersion": 2
    },
    {
      "id": "d974a929-57f3-49de-821b-77a41d46724e",
      "name": "Loop Candidates",
      "type": "n8n-nodes-base.splitInBatches",
      "position": [
        1776,
        360
      ],
      "parameters": {
        "options": {}
      },
      "typeVersion": 3
    },
    {
      "id": "414e3332-8857-4b7a-8f19-4083e60516a1",
      "name": "Prompt 1 \u2014 Score",
      "type": "@n8n/n8n-nodes-langchain.openAi",
      "position": [
        2000,
        288
      ],
      "parameters": {
        "modelId": {
          "__rl": true,
          "mode": "list",
          "value": "gpt-4.1-mini",
          "cachedResultName": "GPT-4.1-MINI"
        },
        "options": {},
        "responses": {
          "values": [
            {
              "role": "system",
              "content": "You are a senior recruitment specialist with 15 years of experience. You evaluate resumes with technical rigor and impartiality. Never inflate scores \u2014 an average candidate should receive an average score. Return ONLY valid JSON, no additional text, no markdown, no backticks."
            },
            {
              "content": "=JOB TEMPLATE:\n{{ JSON.stringify($json.gabarito) }}\n\nCANDIDATE RESUME:\n{{ $json.cv_text }}\n\nSCORING RULES \u2014 follow strictly:\n- 90-100: Meets ALL mandatory requirements with proven, documented experience. Has most differentials.\n- 70-89: Meets most mandatory requirements. Minor gaps compensated by other strengths.\n- 50-69: Partially meets mandatory requirements. Relevant but non-eliminatory gaps.\n- 30-49: Critical gaps in mandatory requirements. Candidate would need significant development.\n- 0-29: Does not meet the minimum requirements for the role.\n\nINSTRUCTIONS:\n- Consider PROVEN experience, not just superficial keyword mentions.\n- Penalize absence of mandatory requirements proportionally to weights in the job template.\n- Value real projects and concrete results over theoretical knowledge.\n- Use the job template weights to calculate the final score AND per-criteria scores.\n\nReturn ONLY valid JSON:\n{\n  \"score\": number between 0 and 100,\n  \"adherence_level\": \"Low|Medium|High|Excellent\",\n  \"justification\": \"3 objective sentences explaining the score. Cite concrete evidence from the resume and specific gaps that justify the rating.\",\n  \"criteria_scores\": {\n    \"mandatory_requirements\": number between 0 and 100,\n    \"differential_requirements\": number between 0 and 100,\n    \"behavioral_competencies\": number between 0 and 100,\n    \"experience\": number between 0 and 100\n  }\n}"
            }
          ]
        },
        "builtInTools": {}
      },
      "credentials": {
        "openAiApi": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.1
    },
    {
      "id": "ab0906db-3de1-4f4e-b749-a680b16d8cd9",
      "name": "Prompt 2 \u2014 Gaps",
      "type": "@n8n/n8n-nodes-langchain.openAi",
      "position": [
        2352,
        288
      ],
      "parameters": {
        "modelId": {
          "__rl": true,
          "mode": "list",
          "value": "gpt-4.1-mini",
          "cachedResultName": "GPT-4.1-MINI"
        },
        "options": {},
        "responses": {
          "values": [
            {
              "role": "system",
              "content": "You are a senior recruitment specialist. Analyze resumes with surgical precision \u2014 cite concrete evidence, not generalities. Return ONLY valid JSON, no additional text, no markdown, no backticks."
            },
            {
              "content": "=JOB TEMPLATE:\n{{ JSON.stringify($('Loop Candidates').item.json.gabarito) }}\n\nCANDIDATE RESUME:\n{{ $('Loop Candidates').item.json.cv_text }}\n\nSCORE RESULT:\n{{ $('Prompt 1 \u2014 Score').item.json.output[0].content[0].text }}\n\nINSTRUCTIONS:\n- For each strength: cite the EXACT evidence from the resume (project, company, technology mentioned).\n- For critical gaps: list only MANDATORY requirements from the template that are absent or insufficiently proven.\n- For secondary gaps: list absent differentials and unevidenced behavioral competencies.\n- Be specific \u2014 \"Experience with n8n in self-hosted production\" is better than \"Automation experience\".\n- Do not repeat information between strengths and gaps.\n\nReturn ONLY valid JSON:\n{\n  \"strengths\": [\"strength with resume evidence \u2014 max 8 items\"],\n  \"critical_gaps\": [\"absent or insufficient mandatory requirement \u2014 be specific\"],\n  \"secondary_gaps\": [\"absent differential or behavioral competency\"]\n}"
            }
          ]
        },
        "builtInTools": {}
      },
      "credentials": {
        "openAiApi": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.1
    },
    {
      "id": "bccbda40-5c02-490b-b2bd-ee1a98ad1e9e",
      "name": "Prompt 3 \u2014 Interview Questions",
      "type": "@n8n/n8n-nodes-langchain.openAi",
      "position": [
        2704,
        288
      ],
      "parameters": {
        "modelId": {
          "__rl": true,
          "mode": "list",
          "value": "gpt-4.1-mini",
          "cachedResultName": "GPT-4.1-MINI"
        },
        "options": {},
        "responses": {
          "values": [
            {
              "role": "system",
              "content": "You are a recruitment specialist with experience in technical and behavioral competency-based interviews. Generate questions that reveal the real candidate, not the prepared one. Return ONLY valid JSON, no additional text, no markdown, no backticks."
            },
            {
              "content": "=JOB TEMPLATE:\n{{ JSON.stringify($('Loop Candidates').item.json.gabarito) }}\n\nCANDIDATE PROFILE:\n- Name: {{ $('Loop Candidates').item.json.candidate_name }}\n- Score: {{ $('Prompt 1 \u2014 Score').item.json.output[0].content[0].text }}\n- Full analysis: {{ $('Prompt 2 \u2014 Gaps').item.json.output[0].content[0].text }}\n\nINSTRUCTIONS:\n- Generate PERSONALIZED questions for this candidate \u2014 not generic HR questions.\n- Prefer situational questions: \"Tell me about a time when...\" or \"How would you handle...\"\n- Avoid yes/no questions.\n- For critical gaps: investigate non-confrontationally \u2014 \"What is your experience with X?\" instead of \"Do you know X?\".\n- For strengths: dig deeper with requests for concrete examples and measurable results.\n- Vary types: at least 2 technical, 2 behavioral, 1 situational.\n- Each question must have a clear and distinct objective.\n\nReturn ONLY valid JSON:\n{\n  \"interview_questions\": [\n    {\n      \"question\": \"full question text\",\n      \"objective\": \"what this question specifically reveals about this candidate\",\n      \"type\": \"technical|behavioral|situational\"\n    }\n  ]\n}\n\nGenerate between 5 and 7 questions."
            }
          ]
        },
        "builtInTools": {}
      },
      "credentials": {
        "openAiApi": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.1
    },
    {
      "id": "01fcae4e-6bcb-48a8-bbd0-8e137e9ef190",
      "name": "Build Analysis Payload",
      "type": "n8n-nodes-base.code",
      "position": [
        3056,
        288
      ],
      "parameters": {
        "jsCode": "const candidate = $('Loop Candidates').item.json;\n\nconst score_data = JSON.parse($('Prompt 1 \u2014 Score').item.json.output[0].content[0].text);\nconst gaps_data  = JSON.parse($('Prompt 2 \u2014 Gaps').item.json.output[0].content[0].text);\nconst pergs_data = JSON.parse($('Prompt 3 \u2014 Interview Questions').item.json.output[0].content[0].text);\n\nreturn [{\n  json: {\n    candidate_id:          candidate.candidate_id,\n    job_id:                candidate.job_id,\n    score:                 score_data.score,\n    nivel_aderencia:       score_data.adherence_level,\n    justificativa_score:   score_data.justification,\n    score_criterios:       score_data.criteria_scores,\n    pontos_fortes:         gaps_data.strengths,\n    gaps_criticos:         gaps_data.critical_gaps,\n    gaps_secundarios:      gaps_data.secondary_gaps,\n    perguntas_entrevista:  pergs_data.interview_questions\n  }\n}];"
      },
      "typeVersion": 2
    },
    {
      "id": "d9ba1c99-fb5d-46f4-8af1-62a9189b4c66",
      "name": "Save Analysis",
      "type": "n8n-nodes-base.postgres",
      "position": [
        3280,
        288
      ],
      "parameters": {
        "query": "INSERT INTO analyses (candidate_id, job_id, score, nivel_aderencia, justificativa_score, score_criterios, pontos_fortes, gaps_criticos, gaps_secundarios, perguntas_entrevista) VALUES ({{ $json.candidate_id }}, {{ $json.job_id }}, {{ $json.score }}, '{{ $json.nivel_aderencia }}', '{{ $json.justificativa_score.replace(/'/g, \"''\") }}', '{{ JSON.stringify($json.score_criterios) }}'::jsonb, '{{ JSON.stringify($json.pontos_fortes) }}'::jsonb, '{{ JSON.stringify($json.gaps_criticos) }}'::jsonb, '{{ JSON.stringify($json.gaps_secundarios) }}'::jsonb, '{{ JSON.stringify($json.perguntas_entrevista) }}'::jsonb) ON CONFLICT (candidate_id) DO UPDATE SET score = EXCLUDED.score, nivel_aderencia = EXCLUDED.nivel_aderencia, justificativa_score = EXCLUDED.justificativa_score, score_criterios = EXCLUDED.score_criterios, pontos_fortes = EXCLUDED.pontos_fortes, gaps_criticos = EXCLUDED.gaps_criticos, gaps_secundarios = EXCLUDED.gaps_secundarios, perguntas_entrevista = EXCLUDED.perguntas_entrevista",
        "options": {},
        "operation": "executeQuery"
      },
      "credentials": {
        "postgres": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.5
    },
    {
      "id": "a74272c7-8c2d-4b3b-a6ac-baa509eb8b12",
      "name": "Update Candidate Status",
      "type": "n8n-nodes-base.postgres",
      "position": [
        3504,
        288
      ],
      "parameters": {
        "query": "UPDATE candidates SET status = 'processed' WHERE id = {{ $('Build Analysis Payload').item.json.candidate_id }}",
        "options": {},
        "operation": "executeQuery"
      },
      "credentials": {
        "postgres": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.5
    },
    {
      "id": "25b0eaf5-2598-48d0-8be9-fb569d37ea03",
      "name": "Check Pending Candidates",
      "type": "n8n-nodes-base.postgres",
      "position": [
        3728,
        288
      ],
      "parameters": {
        "query": "SELECT COUNT(*) AS pending FROM candidates WHERE job_id = {{ $('Build Analysis Payload').item.json.job_id }} AND status = 'pending'",
        "options": {},
        "operation": "executeQuery"
      },
      "credentials": {
        "postgres": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.5
    },
    {
      "id": "3398f657-1294-4e5f-a006-6f8ed5c2e9bc",
      "name": "All Candidates Processed?",
      "type": "n8n-nodes-base.if",
      "position": [
        3952,
        360
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 2,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "id": "cond-pending",
              "operator": {
                "type": "number",
                "operation": "equals"
              },
              "leftValue": "={{ $json.pending }}",
              "rightValue": 0
            }
          ]
        }
      },
      "typeVersion": 2.2
    },
    {
      "id": "397c9184-d234-47d8-bc3a-49087c7874e1",
      "name": "Update Job Status",
      "type": "n8n-nodes-base.postgres",
      "position": [
        4176,
        360
      ],
      "parameters": {
        "query": "UPDATE jobs SET status = 'done' WHERE id = {{ $('Build Analysis Payload').item.json.job_id }}",
        "options": {},
        "operation": "executeQuery"
      },
      "credentials": {
        "postgres": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.5
    },
    {
      "id": "bb43a8a2-8431-47dd-b436-2f71c96430a6",
      "name": "Fetch Full Pool",
      "type": "n8n-nodes-base.postgres",
      "position": [
        4400,
        360
      ],
      "parameters": {
        "query": "SELECT c.name, a.score, a.nivel_aderencia, a.justificativa_score, a.pontos_fortes, a.gaps_criticos FROM candidates c JOIN analyses a ON a.candidate_id = c.id WHERE c.job_id = {{ $('Build Analysis Payload').item.json.job_id }} ORDER BY a.score DESC",
        "options": {},
        "operation": "executeQuery"
      },
      "credentials": {
        "postgres": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.5
    },
    {
      "id": "e4934c9f-1b20-4bbd-b7d8-2512c665f51a",
      "name": "Fetch Job for Summary",
      "type": "n8n-nodes-base.postgres",
      "position": [
        4624,
        360
      ],
      "parameters": {
        "query": "SELECT title, gabarito FROM jobs WHERE id = {{ $('Build Analysis Payload').item.json.job_id }}",
        "options": {},
        "operation": "executeQuery"
      },
      "credentials": {
        "postgres": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.5
    },
    {
      "id": "f8529e9e-1443-4b42-a6c7-c59af19f51c6",
      "name": "Prompt 4 \u2014 Executive Summary",
      "type": "@n8n/n8n-nodes-langchain.openAi",
      "position": [
        4848,
        360
      ],
      "parameters": {
        "modelId": {
          "__rl": true,
          "mode": "list",
          "value": "gpt-4.1-mini",
          "cachedResultName": "GPT-4.1-MINI"
        },
        "options": {},
        "responses": {
          "values": [
            {
              "role": "system",
              "content": "You are a senior HR consultant with experience in executive selection. Analyze candidate pools with strategic vision and objective language. Return ONLY valid JSON, no additional text, no markdown, no backticks."
            },
            {
              "content": "=JOB: {{ $('Fetch Job for Summary').item.json.title }}\n\nJOB TEMPLATE:\n{{ JSON.stringify($('Fetch Job for Summary').item.json.gabarito) }}\n\nANALYZED CANDIDATES (ordered by score):\n{{ JSON.stringify($('Fetch Full Pool').all().map(i => i.json)) }}\n\nAnalyze the complete candidate pool and generate an executive summary for the HR team.\n\nINSTRUCTIONS:\n- Be direct and objective \u2014 HR needs to make a quick decision\n- Clearly identify who to recommend for interview (score >= 70)\n- Point out the most recurring gap in the pool \u2014 this may indicate an issue with the job description or the market\n- The summary should have at most 3 sentences \u2014 concise and executive\n- Do not repeat information already in individual candidate analyses\n\nReturn ONLY valid JSON:\n{\n  \"total_analyzed\": total number of candidates as integer,\n  \"recommended\": [\"Names of candidates with score >= 70\"],\n  \"destaque\": \"1 sentence about the highest scoring candidate and why they stand out\",\n  \"gap_comum\": \"1 sentence about the most recurring gap in the pool\",\n  \"resumo\": \"2-3 sentence executive conclusion with a clear recommendation for next steps\"\n}"
            }
          ]
        },
        "builtInTools": {}
      },
      "credentials": {
        "openAiApi": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.1
    },
    {
      "id": "44192b37-dd99-4a6a-a23b-8138b1b33d28",
      "name": "Build Summary Payload",
      "type": "n8n-nodes-base.code",
      "position": [
        5200,
        360
      ],
      "parameters": {
        "jsCode": "const raw  = $('Prompt 4 \u2014 Executive Summary').item.json.output[0].content[0].text;\nconst data = JSON.parse(raw);\nconst job_id = $('Build Analysis Payload').item.json.job_id;\n\nreturn [{\n  json: {\n    job_id:           job_id,\n    total_analisados: data.total_analyzed,\n    recomendados:     data.recommended,\n    destaque:         data.destaque,\n    gap_comum:        data.gap_comum,\n    resumo:           data.resumo\n  }\n}];"
      },
      "typeVersion": 2
    },
    {
      "id": "221e40ba-5662-45ac-9200-787a460e9c8b",
      "name": "Save Summary",
      "type": "n8n-nodes-base.postgres",
      "position": [
        5424,
        360
      ],
      "parameters": {
        "query": "INSERT INTO job_summaries (job_id, total_analisados, recomendados, destaque, gap_comum, resumo) VALUES ({{ $json.job_id }}, {{ $json.total_analisados }}, '{{ JSON.stringify($json.recomendados) }}'::jsonb, '{{ $json.destaque.replace(/'/g, \"''\") }}', '{{ $json.gap_comum.replace(/'/g, \"''\") }}', '{{ $json.resumo.replace(/'/g, \"''\") }}') ON CONFLICT (job_id) DO UPDATE SET total_analisados = EXCLUDED.total_analisados, recomendados = EXCLUDED.recomendados, destaque = EXCLUDED.destaque, gap_comum = EXCLUDED.gap_comum, resumo = EXCLUDED.resumo",
        "options": {},
        "operation": "executeQuery"
      },
      "credentials": {
        "postgres": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 2.5
    }
  ],
  "active": false,
  "settings": {
    "executionOrder": "v1"
  },
  "versionId": "dc533c0c-4541-411b-a15d-81cd4c668502",
  "connections": {
    "Receive CVs": {
      "main": [
        [
          {
            "node": "Fetch Job and Candidates",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Save Analysis": {
      "main": [
        [
          {
            "node": "Update Candidate Status",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Fetch Full Pool": {
      "main": [
        [
          {
            "node": "Fetch Job for Summary",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Loop Candidates": {
      "main": [
        [],
        [
          {
            "node": "Prompt 1 \u2014 Score",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Prompt 2 \u2014 Gaps": {
      "main": [
        [
          {
            "node": "Prompt 3 \u2014 Interview Questions",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Save Job Template": {
      "main": [
        [
          {
            "node": "Prepare Candidates",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Update Job Status": {
      "main": [
        [
          {
            "node": "Fetch Full Pool",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Prepare Candidates": {
      "main": [
        [
          {
            "node": "Loop Candidates",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Prompt 1 \u2014 Score": {
      "main": [
        [
          {
            "node": "Prompt 2 \u2014 Gaps",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Job Template exists?": {
      "main": [
        [
          {
            "node": "Prepare Candidates",
            "type": "main",
            "index": 0
          }
        ],
        [
          {
            "node": "Prompt 0 \u2014 Extract Job Template",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Build Summary Payload": {
      "main": [
        [
          {
            "node": "Save Summary",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Fetch Job for Summary": {
      "main": [
        [
          {
            "node": "Prompt 4 \u2014 Executive Summary",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Build Analysis Payload": {
      "main": [
        [
          {
            "node": "Save Analysis",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Update Candidate Status": {
      "main": [
        [
          {
            "node": "Check Pending Candidates",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Check Pending Candidates": {
      "main": [
        [
          {
            "node": "All Candidates Processed?",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Fetch Job and Candidates": {
      "main": [
        [
          {
            "node": "Job Template exists?",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "All Candidates Processed?": {
      "main": [
        [
          {
            "node": "Update Job Status",
            "type": "main",
            "index": 0
          }
        ],
        [
          {
            "node": "Loop Candidates",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Prompt 4 \u2014 Executive Summary": {
      "main": [
        [
          {
            "node": "Build Summary Payload",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Prompt 3 \u2014 Interview Questions": {
      "main": [
        [
          {
            "node": "Build Analysis Payload",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Prompt 0 \u2014 Extract Job Template": {
      "main": [
        [
          {
            "node": "Save Job Template",
            "type": "main",
            "index": 0
          }
        ]
      ]
    }
  }
}

Credentials you'll need

Each integration node will prompt for credentials when you import. We strip credential IDs before publishing — you'll add your own.

• openAiApi
• postgres

About this workflow

Webhook receives a job ID and list of candidate IDs from your database If the job has no template yet, Prompt 0 reads the job description and automatically extracts mandatory requirements, differentials, behavioral competencies and sets the weight of each criterion For each…

Source: https://n8n.io/workflows/13876/ — original creator credit. Request a take-down →

More AI & RAG workflows → · Browse all categories →

Related workflows

Workflows that share integrations, category, or trigger type with this one. All free to copy and import.