{"id":49585791,"url":"https://github.com/jexp/neo4j-getting-started-skill","last_synced_at":"2026-05-03T22:35:09.182Z","repository":{"id":352753275,"uuid":"1215841265","full_name":"jexp/neo4j-getting-started-skill","owner":"jexp","description":"An agent skill to get from zero to success with agentic coding environment with Neo4j (locally or on Aura)","archived":false,"fork":false,"pushed_at":"2026-04-21T00:30:38.000Z","size":238,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-04-21T02:38:36.225Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/jexp.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-04-20T09:59:09.000Z","updated_at":"2026-04-21T00:30:43.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/jexp/neo4j-getting-started-skill","commit_stats":null,"previous_names":["jexp/neo4j-getting-started-skill"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/jexp/neo4j-getting-started-skill","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jexp%2Fneo4j-getting-started-skill","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jexp%2Fneo4j-getting-started-skill/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jexp%2Fneo4j-getting-started-skill/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jexp%2Fneo4j-getting-started-skill/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/jexp","download_url":"https://codeload.github.com/jexp/neo4j-getting-started-skill/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jexp%2Fneo4j-getting-started-skill/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32587821,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-03T22:12:39.696Z","status":"ssl_error","status_checked_at":"2026-05-03T22:09:10.534Z","response_time":103,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2026-05-03T22:35:08.600Z","updated_at":"2026-05-03T22:35:09.167Z","avatar_url":"https://github.com/jexp.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# neo4j-getting-started-skill\n\nA Claude Code skill that guides a user or autonomous agent from **zero to a working Neo4j application** — in ≤15 minutes (autonomous) or ≤90 minutes (human-in-the-loop).\n\n## What it does\n\nThe skill executes 8 ordered stages, each backed by a focused reference file:\n\n```\n0-prerequisites → 1-context → 2-provision → 3-model → 4-load\n                → 5-explore → 6-query → 7-build\n```\n\nAt the end the user has:\n- A running Neo4j database (Aura Free, Aura Pro, or Docker)\n- A schema with constraints and indexes\n- Data loaded (demo, synthetic, CSV, or documents)\n- A Cypher query library validated against the live DB\n- A runnable application (Jupyter notebook, Streamlit app, FastAPI backend, GraphRAG pipeline, or MCP config)\n- A browser URL to see the graph visually\n\n## Repository layout\n\n```\nneo4j-getting-started-skill/   ← the skill (install this)\n  SKILL.md                     ← orchestrator (8 stages, 7 success gates)\n  AGENTS.md                    ← conventions for agents building/modifying this skill\n  references/\n    0-prerequisites.md … 7-build.md   ← one file per stage\n    capabilities/\n      execute-cypher.md        ← 3 execution methods (MCP, cypher-shell, HTTP)\n      cypher-authoring.md      ← authoring rules + pitfalls table\n      mcp-config.md            ← neo4j-mcp config for Claude Desktop / Claude Code\n      kg-from-documents.md     ← GraphRAG ingestion pipeline\n    domain-patterns.md         ← graph model templates by domain\nneo4j-getting-started-skill-tests/\n  personas/                  ← YAML personas (Alex, Sam, …)\n  fixtures/                  ← Pre-populated work dirs for mid-flow testing\n  harness/\n    runner.py                ← integration test runner\n    validator.py             ← 7-gate validation\n  results/                   ← test run JSON output (git-ignored)\n\nvalidate_cypher.py             ← syntax-validates all Cypher blocks in references/\ntest_load_scripts.py           ← validates Python load scripts against a live DB\naura.env                       ← Aura API credentials (git-ignored)\n```\n\n## Key design decisions\n\n| Decision | Rationale |\n|----------|-----------|\n| SKILL.md is a thin orchestrator | Agent reads only the active stage reference; avoids \u003e700-line system prompts |\n| Stages have `progress.md` tracking | Enables resumability — if a run fails mid-flow, the next run picks up where it left off |\n| Preferred load: DataFrame + `$rows` batches | Works on Aura (no file:// access); any data source via pandas |\n| `CALL {} IN TRANSACTIONS` needs `session.run()` | `execute_query()` uses managed transactions which reject this clause |\n| CYPHER 25 + QPP syntax throughout | Neo4j 2026.01+; `(()-[:R]-\u003e()){0,5}` replaces variable-length `[:R*0..5]` |\n| Aura provisioning via direct REST (not CLI) | No extra binary needed; curl works everywhere |\n| `graph_visible` is a hard success gate | Seeing nodes+edges rendered is the \"it clicks\" moment; never skip |\n| v1 Python only | Focus on quality; JavaScript in phase 2 |\n\n## Install the skill\n\n```bash\nmake install-skill\n# installs to ~/.claude/skills/neo4j-getting-started-skill/\n```\n\nThen trigger from any Claude Code session:\n```\n/neo4j-getting-started-skill\n```\n\n## Integration testing\n\nIntegration tests run Claude Code autonomously with a persona YAML and validate the produced files and database state against 7 success gates.\n\n**Prerequisites:**\n\n1. Copy `aura.env.example` → `aura.env` and add your credentials:\n   ```\n   CLIENT_ID=...\n   CLIENT_SECRET=...\n   PROJECT_ID=...          # recommended for multi-project accounts\n   ORGANIZATION_ID=...     # recommended for multi-org accounts\n   NEO4J_URI=...           # optional — pre-populates .env, skips provisioning\n   NEO4J_PASSWORD=...\n   ```\n2. Install Python deps: `pip install python-dotenv pyyaml neo4j`\n\n**Run tests:**\n\n```bash\n# Full end-to-end (provisions a real Aura instance)\nmake integration-alex\nmake integration-sam\n\n# Start from stage 5-explore (stages 0-4 pre-populated from fixture)\nmake integration-alex-from-explore\n\n# Validate all Cypher blocks in references/ (requires localhost:7687, db=jtbd)\nmake test-cypher\n\n# Validate Python load scripts\nmake test-load\n\n# Check skill file structure (no Claude, no DB)\nmake test-quick\n```\n\n**How the runner works:**\n- Creates a fresh `tempdir` for each run — no source directory pollution\n- Copies `aura.env` into the work dir; if it contains `NEO4J_URI`+`NEO4J_PASSWORD`, writes `.env` automatically (skips provisioning stage)\n- Installs the skill to `~/.claude/skills/`, runs Claude via `--append-system-prompt`, uninstalls on exit\n- Streams Claude's output live; prints `[tool]` lines for each tool call\n- After the run, validates all 7 success gates against the work dir and live DB\n\n**Fixture-based testing** (`neo4j-getting-started-skill-tests/fixtures/alex_after_load/`):\nPre-populates stages 0-4 so the test only exercises stages 5-7.\nUsed by `make integration-alex-from-explore`. The `__NEO4J_URI__` placeholder in `progress.md` is substituted at runtime from `aura.env`.\n\n## Success gates\n\n| Gate | Stage | Condition |\n|------|-------|-----------|\n| `db_running` | provision | `driver.verify_connectivity()` succeeds |\n| `model_valid` | model | ≥2 node labels, ≥1 rel type, ≥1 constraint in DB |\n| `data_present` | load | `MATCH (n) RETURN count(n)` ≥ 50 |\n| `queries_work` | query | ≥5 queries; ≥2 traversals; ≥3 return ≥1 result |\n| `graph_visible` | explore | Browser URL or notebook viz delivered |\n| `app_generated` | build | Artifact exists, passes syntax, returns non-empty results |\n| `integration_ready` | build | MCP config present (if requested) |\n\n## Assumptions\n\n- Claude Code is available as `claude` on PATH\n- Python ≥3.10\n- `neo4j-mcp` binary is downloaded during the `prerequisites` stage (or pre-installed)\n- Aura Free has a 1-3 min provisioning wait; this does not count against the 15-min budget\n- GDS (Graph Data Science) is not available on Aura Free — the skill guards all GDS queries\n- `CALL {} IN TRANSACTIONS` requires `session.run()`, not `execute_query()` — validated in test harness\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjexp%2Fneo4j-getting-started-skill","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fjexp%2Fneo4j-getting-started-skill","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjexp%2Fneo4j-getting-started-skill/lists"}