li-dm

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: the skill only writes drafts — it never connects to LinkedIn, never logs in, and never sends anything. To actually send the messages it writes I need my own LinkedIn account, and I send every message by hand myself (see the use-time disclosure in point 2: the only thing that ever reaches LinkedIn is a message I paste and send myself, from my own account). At use time I also supply the specifics about the person myself, in our conversation: their name, role and company, the specific hook (a post they wrote, something their company shipped, a mutual connection, a talk they gave), and what I want (a conversation, a referral, a job, a sale). One part of the pack workflow is NOT included in this install: SKILL.md says the output is "All humanized through /li-human", and /li-human is a separate skill elsewhere in the same repository that this install does not copy — disclosed in point 2. Install "li-dm" for me. It writes LinkedIn outreach that is not a pitch: first it asks me, in one batched question, for the who, the hook and what I want, and it says so if there is no specific reason to message this person today. It then writes (1) a connection invite note under 200 characters including spaces — one specific reference to them, one line of who I am, no ask at all — with the character count shown; (2) a first message to send a day after they accept, two to four sentences, referencing the same specific thing, giving something (a number, a template, a name, an answer) before asking, with one small ask and no calendar link; (3) exactly two follow-ups: one at +4 days that adds something new (never "just bumping this"), and a close-the-loop one at +10 days that says I will stop — then it stops; a third follow-up is never written. Its hard rules: never fabricate a mutual connection, a shared school or having read something I have not read; never open with "I hope this message finds you well"; never an automated connection or message sequence (SKILL.md says automated outreach tools violate LinkedIn's User Agreement and get accounts restricted); and never more than 20 invites a day. Repository: https://github.com/avi691/linkedin-agent-skill (the skill is the whole skills/li-dm/ folder at the pinned commit: SKILL.md alone, 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-dm/ 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 14b0ee4cffedc2d3eb1d9f8dd6d6cce7bc04c95e221ea1e68809fb5aad525747 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 (3,117 bytes). There are no scripts, no code, no data files and nothing executable in the folder: it is a writing method in Markdown that you follow in our conversation. - Dependency outside the folder, disclosed here: the Output section of SKILL.md ends with "All humanized through /li-human. The user sends every one of them by hand." /li-human is a different skill in the same repository (the skills/li-human/ folder, which carries its own SKILL.md plus humanize.py, detect.py and slop.json) and it is NOT part of this install and NOT in the fingerprint above. So with this install alone, the /li-human humanizing pass named in SKILL.md is not available: never claim you ran it. If I have not separately installed /li-human, write the drafts following li-dm's own rules and tell me plainly that the /li-human pass was not run. Do not install /li-human, any of the other 10 skills in the repository, or anything else during this install. - Pack-level voice file, also outside this install: the repository README and templates/voice.md say every skill in the pack reads a voice file copied to ~/.claude/linkedin/voice.md — but li-dm's own SKILL.md never names voice.md or that path; it works from the specifics I give you in the conversation (who, hook, what I want). Do not create, copy or fill in voice.md during this install; setting up the pack's voice file is a separate, optional decision of mine. - 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 no script in the folder), no key or account is asked for, there is no telemetry in the folder, and nothing in SKILL.md reads cookies, a LinkedIn session, a browser store or any credential: the file contains no API, no URL to call and no login step at all. - What leaves this machine at use time, disclosed here before it ever happens — using the skill is a separate step that only starts when I ask you to write outreach to a specific person: the skill itself sends nothing and connects to nothing. I give you the recipient's name, role, company, the hook and what I want, here in our conversation; you return drafts here in our conversation (the invite note with its character count, the first message, and the two follow-ups labelled with the day they go out). Then I, by hand, from my own LinkedIn account, paste and send each message on LinkedIn: only then does the message text I approved reach LinkedIn, addressed to that recipient, under my own account and LinkedIn's own terms — and that manual sending is my action, not the skill's. Nothing is ever posted or sent automatically, no sequence is ever run, and there is no other recipient: no third-party service, no detector API, no analytics. - The sending discipline from SKILL.md is part of the skill, and you must hand it back to me with the drafts: wait a day after acceptance before the first message; follow-up 1 at +4 days only if it adds something new, follow-up 2 at +10 days, then stop for good; no more than 20 invites a day; never fabricate a connection, school or reading; no calendar link in message one. - Do not log in to LinkedIn, do not open LinkedIn in a browser, do not send or schedule any message, and do not write any outreach during this install: using the skill is my decision, later, when I actually name a person. 3. When I say "yes, install", install it as a skill: copy the whole li-dm folder into the agent's skills directory, in a folder named "li-dm", keeping its 1 file (SKILL.md at its root). A manual clone followed by copying the skills/li-dm 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-dm), the 1 file in place with matching fingerprint. There is no script to run to verify. 5. Report what was installed, where, and what I still need to do myself (for example: give you the specifics for a person — name, role, company, the real hook and what I want — when I ask for outreach; know that the /li-human pass named in SKILL.md is not part of this install unless I install that separate skill myself; send every message by hand from my own LinkedIn account, respecting the waits and the stop rule above; using the skill is a separate step that only starts when I ask for it, and the only use-time send disclosed above — my own manual paste-and-send on LinkedIn — happens only then, by my hand). 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.