li-carousel

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: none for the install itself: no account, no API key, no paid service, and nothing else to install first. For using the skill afterwards, read this first: it produces copy and a PDF locally and uploads nothing — I post the finished PDF on LinkedIn myself, under my own account, only if I choose to. Building the PDF relies on a tool that is NOT in the skill folder and is NOT installed by this install: Chrome headless (--print-to-pdf) or any HTML-to-PDF tool I already use, as SKILL.md puts it. Two steps in SKILL.md also point to sibling skills that are NOT included in this install: running the copy through /li-human (a separate skill in the same repository; if I have it installed, its bundled Python scripts run locally on my text) and, if my idea is a single claim rather than a sequence, being handed to /li-post instead. If I have a brand skill or design system in the project, the skill uses it for the PDF palette rather than inventing one. I also provide the raw material myself: the idea or list to turn into a carousel, my LinkedIn handle (which goes small in the bottom corner of every slide), and my approval of the copy before any PDF is built. Install "li-carousel" for me. It builds a LinkedIn document post (carousel): slide-by-slide copy for 8-12 slides — a cover (a hook of 6 words or fewer plus one line of promise), a stake slide (why this matters, in one sentence), one idea per slide (a headline of 3-7 words and at most 25 words under it, numbered like 3/10), a recap slide that lists the whole thing so a screenshot of it is useful on its own, and a last slide with a single call to action (follow, comment a keyword, or the link — one only) — then the 2-3 lines of post text that sit above the carousel in the feed, a pass through /li-human, and, only after I approve the copy, a PDF built as HTML (one page per slide, 1080x1350, i.e. 4:5) to upload to LinkedIn myself. It refuses the wrong shape: if the idea is one claim with no sequence, it says so and hands me to /li-post for a text post instead of splitting one claim across eight slides. Repository: https://github.com/avi691/linkedin-agent-skill (the skill is the whole skills/li-carousel/ folder at the pinned commit: SKILL.md only, 1 file) Pinned version: install the code at commit e4ec4fb6f1b0e227eb9b14b31b735072db5df9cb exactly. Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Clone https://github.com/avi691/linkedin-agent-skill into a temporary folder, check out commit e4ec4fb6f1b0e227eb9b14b31b735072db5df9cb, and copy the skills/li-carousel/ folder from that checkout. Then compare every file in the folder with the fingerprints recorded for this listing on 2026-10-19: [FINGERPRINTS] - SKILL.md: SHA-256 fa1a8801e661ed83e55eb5c44e0810a773b490eac930bd3200d783d4bbd23094 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, which services it connects to, whether it reads or sends any personal information or keys, and anything that seems unusual. Stop on any red flag and tell me before going further. Confirm these points from the file: - The skill folder holds 1 file only: SKILL.md (2,860 bytes). There is no script, no data file, and no bundled PDF tool in the folder. The copy workflow, the slide structure and the PDF specification are all described in that single file. - Dependencies outside the folder, exactly as written in SKILL.md: (a) the PDF step — SKILL.md says to build the deck as HTML (one <section> per slide, width:1080px; height:1350px; page-break-after:always, a single accent colour, type no smaller than 28px) and print it to PDF with "Chrome headless --print-to-pdf, or any HTML-to-PDF you already use", so it relies on Chrome or another HTML-to-PDF tool already on my machine; LinkedIn's stated limits in the file are 1080x1350 (4:5), under 100MB and under 300 pages. (b) The humanizing step — SKILL.md says "Run both through /li-human" (both the slide copy and the post text): /li-human is a sibling skill in the same repository (skills/li-human/, with its own humanize.py, detect.py and slop.json), it is NOT part of this install, and if it is not installed that step cannot run as named — tell me so rather than silently skipping or inventing it. (c) The handoff — if my idea is one claim, SKILL.md says to hand me to /li-post, another sibling skill, also not included. (d) An optional brand skill or design system already in my project: "If the user has a brand skill or design system in this project, use it and do not invent a palette." SKILL.md itself does not read templates/voice.md or any voice file, does not read any other file in the repository, and needs no other of the 11 skills in the pack to write the slide copy and specify the PDF. - Nothing is sent anywhere during installation itself; the only download at install time is the skill folder from GitHub at the pinned commit. No script is run during installation, no key is asked for, and there is no telemetry in the folder. - No LinkedIn access by the skill, at install time or at use time: SKILL.md contains no URL, no API call, and no instruction to log in, and it never touches LinkedIn cookies, a LinkedIn session, a password, an API key, a token or any credential file. Its last line is explicit: "Nothing is uploaded to LinkedIn. The user posts the PDF themselves." What leaves this machine at use time, disclosed here before it ever happens: nothing is sent by the skill itself. Writing the copy and building the HTML/PDF are local, between you and me, using the idea and handle I provide in our conversation. The only outward step is mine, separate from using the skill and only when I decide to publish: I upload the finished PDF and paste the post text on LinkedIn myself, under my own account — that upload sends exactly the PDF and post text I approved to LinkedIn. If I have /li-human installed separately and the copy is run through it, its own SKILL.md describes local Python scripts that run on my machine on my text; it is outside this install's perimeter, so read its files before running it and report what you find there separately. - Do not install Chrome or any HTML-to-PDF tool during this install, do not run /li-human or /li-post, do not write any HTML, do not build any PDF, and do not open or post anything on LinkedIn: PDF tooling, the sibling skills and publishing are my decisions, later, when I actually ask for a carousel. 3. When I say "yes, install", install it as a skill: copy the whole li-carousel folder into the agent's skills directory, in a folder named "li-carousel", keeping its 1 file (SKILL.md at its root). A manual clone followed by copying the skills/li-carousel folder into ~/.claude/skills/ is the route described here; any other route is fine only as long as every installed file matches its fingerprint. 4. Verify with no network calls: frontmatter valid (name: li-carousel), the 1 file in place with a matching fingerprint. Do not build a PDF or run any sibling skill to verify. 5. Report what was installed, where, and what I still need to do myself (for example: provide the idea or list for a carousel and my LinkedIn handle; approve the slide copy and post text before any PDF is built; have Chrome headless or another HTML-to-PDF tool available myself if I want the PDF built; install /li-human separately myself if I want the humanizing pass SKILL.md calls for, and expect a handoff suggestion to /li-post if my idea is a single claim; post the finished PDF on LinkedIn myself, since the skill uploads nothing; using the skill is a separate step that only starts when I ask for it, and the only outward send — my own manual upload to LinkedIn of the PDF and text I approved — happens only then, by me). 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.