li-plan

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, and none for using it either: no LinkedIn account to connect, no API key, no paid service, no signup, and nothing else to install first. The pack's README says "Free, MIT, no signup, no API key, nothing to connect", and this skill is a single Markdown file with no code and no scripts. Read this first for using it afterwards: the skill works entirely in conversation and in local files — it reads, if they already exist, my local files ~/.claude/linkedin/voice.md and ~/.claude/linkedin/log.md, asks me four things if they do not (what I sell and to whom, the three or four themes I want to be known for, what actually happened this week, and ten to twenty people or companies worth being visible to), and writes the finished plan to ~/.claude/linkedin/plan.md so the other skills can read it. It posts nothing, schedules nothing, connects to no LinkedIn account, API or browser, reads no cookie or session, and sends nothing anywhere (all of this is disclosed in point 2). One incomplete outside dependency to know up front: for each slot the skill names a hook formula number from li-post/hooks.json — that file (21 formulas) belongs to the sibling skill li-post, it is NOT inside this skill's folder, and this install does NOT copy it; without li-post installed separately, the numbered hook formulas cannot be looked up. Install "li-plan" for me. It is the weekly planning skill — "the control room" — of a LinkedIn skill pack: it decides what gets executed before anything else runs, and you run it once a week, on the same day. It builds my LinkedIn week: what to post, when to post it, and who to engage with. Four posts a week beats seven; the mix never puts two of the same type back to back — Proof 1 per week (something that happened, with a number), Opinion 1 per week (a position that could lose me followers), Teach 1 per week (one thing the reader can do today), Story 1 per fortnight (a scene with dialogue and a cost), Offer 1 per fortnight (what I sell, said plainly, no apology). For each slot it gives the theme, the specific angle drawn from what actually happened this week (not a topic — "AI" is not a plan), and the hook formula number from li-post/hooks.json that fits it. Posting times: for a B2B audience in one timezone, Tuesday to Thursday, 7:30–9:30am local is the working default, Monday afternoon and Friday morning are the second tier, weekends are for personal-story posts or nothing, times are anchored to my audience's timezone rather than mine if they differ — and SKILL.md says plainly that the day and hour matter far less than whether the first line is good. Engagement: a list of 10, split 5 reach (people with an audience I want), 3 peers (same level, same field), 2 buyers (people who could actually buy, commented on for weeks before any DM, never pitched in a comment), at 20 minutes a day, before posting, not after. Its output is a week grid (for example MON engage only, TUE 8:15am PROOF with its hook number and angle), the ENGAGE list, and the line "Say 'write Tuesday' and I will draft it." The plan avoids repeating a theme from the last fortnight by reading voice.md and log.md when they exist. Nothing is scheduled or posted anywhere — this is a plan, and I run it. Repository: https://github.com/avi691/linkedin-agent-skill (the skill is the whole skills/li-plan/ 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-plan/ 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 5799665246d1c47ab6b37840bd4f4f94700f24ff54ae4d3a1fbaf49b1cd71f9b 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 (3,617 bytes). There is no code, no script, and nothing executable in it. The repository at the pinned commit holds 22 tracked files in total — the other 10 li-* skills (li-audit, li-carousel, li-comment, li-dm, li-human, li-inbox, li-post, li-profile, li-reply, li-repurpose), templates/voice.md, the README, the LICENSE and the plugin manifests — and none of those is part of this install: copy the skills/li-plan/ folder only. - 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 (there is none), no key or account is asked for, and there is no telemetry in the folder (there is no code at all). - What happens at use time, disclosed here before it ever happens — using the skill is a separate step that only starts when I ask to plan my week: the skill itself makes no network call of any kind. It does not log into LinkedIn, it reads no LinkedIn cookie or session, it uses no LinkedIn API and no browser automation, and it posts and schedules nothing: SKILL.md ends with "Nothing is scheduled or posted anywhere - this is a plan, and the user runs it", and the pack's README says the skills do not post to LinkedIn and should not. Everything the skill works with is information I give it in conversation: what I sell and to whom, my three or four themes, what actually happened this week (a client call, a number, a mistake, a thing I built, an argument I had), and ten to twenty people or companies worth being visible to. That information stays in our conversation and in the local files named below; the skill sends it nowhere. - The local files at use time: the skill reads ~/.claude/linkedin/voice.md and ~/.claude/linkedin/log.md if they exist, so the plan does not repeat a theme from the last fortnight — those are my own local files (the voice.md template in the repository, templates/voice.md, is a form to fill in and copy there by hand; it is NOT installed or copied by this install). If voice.md and log.md do not exist, the skill asks me the four things listed above and writes them down — SKILL.md does not name the file those notes go to, so if you write them somewhere, tell me where. The skill then writes the finished plan to ~/.claude/linkedin/plan.md "so the other skills can read it"; that is a local file write on my machine, disclosed here, and it happens only when I use the skill, never during installation. - The outside dependency, stated factually: SKILL.md tells it, for each slot, to give "the hook formula number from li-post/hooks.json that fits it". That file is skills/li-post/hooks.json in the same repository (21 hook formulas, 9,414 bytes at the pinned commit) — it belongs to the sibling skill li-post, it is not in the li-plan folder, and this install does not copy li-post or hooks.json. If li-post is not already installed on my machine, you cannot look those formulas up: do not invent formula numbers — tell me the numbered hook lookup is unavailable without the sibling skill, give me the theme and the specific angle for each slot anyway, and let me decide separately whether I want li-post installed. li-plan's own text never invokes li-human or any other sibling skill; its output example ends with "Say 'write Tuesday' and I will draft it", and actually drafting a post later is a separate step I ask for, not part of this planning skill's install or run. - Do not create ~/.claude/linkedin/voice.md, log.md or plan.md during this install, do not ask me the four planning questions, and do not plan a week: using the skill is my decision, later, when I actually ask for a weekly plan. 3. When I say "yes, install", install it as a skill: copy the whole li-plan folder into the agent's skills directory, in a folder named "li-plan", keeping its 1 file (SKILL.md at its root). A manual clone followed by copying the skills/li-plan 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-plan), the 1 file in place with a matching fingerprint. There is no script to run for verification — do not run anything. 5. Report what was installed, where, and what I still need to do myself (for example: nothing to set up — no account, no key, nothing to connect; when I want to use it, ask for a weekly plan and be ready to give the four inputs, or prepare ~/.claude/linkedin/voice.md myself first from the repository's templates/voice.md if I want my voice profile read; know that the finished plan will be written to ~/.claude/linkedin/plan.md, that the numbered hook formulas come from the sibling li-post/hooks.json which this install did not install, and that nothing will ever be posted or scheduled by this skill — posting stays manual, by me; using the skill is a separate step that only starts when I ask for it, and the local reads and writes disclosed above happen only then). 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.