Graphify

Before you start: read this whole install prompt, tell me in a few lines what this skill does and what installing it involves, and wait for my go-ahead before you do anything else. Prerequisites — Reinforced framing, disclosed up front: installing this listing only copies 9 text files (one SKILL.md and 8 reference files) and runs nothing. The wall is at use time, not at install time: the skill drives the separate graphifyy tool, and if that tool is not already installed, the skill's own first step installs it (uv tool install, pipx, or pip install graphifyy) when you first use it — not during this install. Also at use time, and only on the flags or inputs you give it: it can clone repositories you pass as URLs, transcribe video or audio with Whisper, use a Gemini key for semantic extraction if you have set GEMINI_API_KEY or GOOGLE_API_KEY (the skill itself states graphify needs no API key and must never ask you for one), and push the graph to a local Neo4j or FalkorDB instance. Upstream is dual-licensed: the LICENSE file is Apache-2.0 and LICENSE-MIT is MIT. Install "Graphify" for me. It turns a folder of files — code, docs, SQL, PDFs, images, even video — into a persistent, queryable knowledge graph: structural extraction from code plus semantic extraction from content, community detection, and three outputs (an interactive HTML view, GraphRAG-ready JSON, and a plain-language GRAPH_REPORT.md), with query, path and explain flows against the built graph. This listing installs the Claude variant exactly as the upstream installer does: upstream's install code (graphify/install.py, platform "claude") copies the packaged graphify/skill.md to .claude/skills/graphify/SKILL.md and installs the references sidecar from graphify/skills/claude/references/ next to it. The other skill-<harness>.md variants and the other skills/<harness>/ folders in the repository are generated copies of the same content for other agent harnesses; they are deliberately excluded here (one canonical skill only). Repository: https://github.com/Graphify-Labs/graphify (the Claude install scope at the pinned commit is 9 files: graphify/skill.md, installed under the name SKILL.md, plus the 8 files of graphify/skills/claude/references/, installed under references/) Pinned version: install the code at commit c588b72d936927790a87ad009955078ed40562a3 exactly. Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Clone https://github.com/Graphify-Labs/graphify into a temporary folder and check out commit c588b72d936927790a87ad009955078ed40562a3. The source-to-installed mapping is: graphify/skill.md becomes SKILL.md, and each file of graphify/skills/claude/references/ keeps its name under references/. Compare every installed file with the fingerprints recorded for this listing on 2026-10-22, listed here under their installed names: [FINGERPRINTS] - SKILL.md: SHA-256 75ce4bddb9e118a9d72ded3e7f8281d670943dce853ac5886f2b99a2d8726e5b - references/add-watch.md: SHA-256 1e77351815547334c7e030f018dba513b303d397bf73d11fac47df03ace3b3f4 - references/exports.md: SHA-256 c39a2f8062be9a5ac319220d6e66ca586ed0b0798292b1321206834be74c1b33 - references/extraction-spec.md: SHA-256 214c474186ed26f47fafd5be941b29c3524d5e8a362814587f1c5d9883795d7c - references/github-and-merge.md: SHA-256 e5ebd90c7686f50363ff7a535556bc2f596d4c47ec1e6c8b95e11e36a0dfea2b - references/hooks.md: SHA-256 b9a4e9f66813c6fc720589f1071d1a03c95756ab7101447e14e57291fe7844e5 - references/query.md: SHA-256 e99011beeff0cf40dc1775b1e0cd2eab0e5f950deeeb8353ae01cbe57adbff00 - references/transcribe.md: SHA-256 676a1e39aa6d43cdfcc416ec56616e36f7bad74066a82bd33a9485515b9a865c - references/update.md: SHA-256 c76b855ef9380d766b1a0b271b49bc7b40cf5b5b72558be1a021f99f9a9ea1cc If a file does not match, it changed since the listing was prepared: do not install it yet, tell me what you found and let me decide. 2. Read SKILL.md in full, and tell me in plain words: what the skill does when used, what it connects to, whether it reads or sends any personal information or credentials, and anything that seems unusual. Stop on any red flag and tell me before going further. Confirm these points from the files: - The installed scope is 9 files only: SKILL.md (47,544 bytes, frontmatter name graphify, the packaged graphify/skill.md) and the 8 references/*.md files (add-watch, exports, extraction-spec, github-and-merge, hooks, query, transcribe, update). The Python package of the graphifyy tool is NOT part of this listing and is not installed by it. - Nothing is sent anywhere and nothing is installed during installation itself; the only download at install time is the repository from GitHub at the pinned commit. - At use time the skill runs the graphifyy tool on folders or repositories I give it, locally: it can clone a repository I pass as a URL, transcribe video or audio with Whisper, optionally use a Gemini key I have set for semantic extraction (it must never ask me for a key), and push to a local Neo4j or FalkorDB only on the matching flags. None of that happens during this install. - One comment in SKILL.md explains writing a JSON sidecar from Python instead of a shell redirect, so the same block renders on PowerShell hosts without console encoding drift; it is a comment about output encoding, not decoded or executed code. - Do not build any graph, do not install graphifyy, and do not use the skill in any way during this install: using it is a separate step that only starts when I ask for it later. 3. When I say "yes, install", install it as a skill: in the agent's skills directory, create a folder named exactly "graphify", write SKILL.md at its root and the 8 reference files under references/, keeping the fingerprints above. 4. Verify with no network calls: frontmatter valid (name: graphify), the 9 files in place with matching fingerprints. Do not run anything from the folder and do not use the skill to verify. 5. Report what was installed, where, and what I still need to do myself (for example: that the graphifyy tool is installed separately at first use, that a Gemini key is optional and never required, and that using the skill is a separate step that only starts when I ask for it later). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.