HarshithNayakaL

LinkedIn Algorithm Source — Development skill for Claude Code

Development community

linkedin-boost — Claude skill for every LinkedIn surface — feed, search, recruiter sourcing, applicant ranking — built from LinkedIn Engineering's published mechanism, and explicit about which popular.

How to install LinkedIn Algorithm Source

This entry records only its repository, not the path inside it, so there is no exact command to give. Open HarshithNayakaL/linkedin-algorithm-source and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What LinkedIn Algorithm Source does

linkedin-boost — Claude skill for every LinkedIn surface — feed, search, recruiter sourcing, applicant ranking — built from LinkedIn Engineering's published mechanism, and explicit about which popular rules are folklore.

Alternatives in Development

  • JWT Scan — JWT attack toolkit (offline) — alg:none forgery, RS256→HS256 algorithm confusion, weak-secret crack, static cl 4.5k ★
  • Godogen Source Repo — This repository contains two source trees for the same project: - claude/ — Claude Code source tree - codex/ 2.8k ★
  • Deep Research From Scratch — Deep research has broken out as one of the most popular agent applications 768 ★

README

linkedin-boost

A Claude skill for **anything LinkedIn** — posting, profile, job hunting, hiring, network growth, sales — built from **LinkedIn Engineering's own published descriptions** of how its ranked surfaces work, not from growth-hack blogs.

Four command-line drivers plus an agent playbook. No dependencies.

LinkedIn is not one algorithm

This is the framing the whole skill is built on. LinkedIn is **several independently-ranked surfaces**, each with its own retrieval, inputs and failure mode — and most bad advice comes from treating them as one thing.

Surface Owns Source
Feed post reach, comments, dwell Mar 2026 · Jun 2021 · May 2020
Follow recommendations follower growth Aug 2026
People / Job Search being found Jan 2026
Recruiter sourcing being approached by recruiters Oct 2024
Applicant ranking hearing back after applying Jul 2020
People You May Know network growth Jun 2024
InMail, notifications, Groups, SSI — not published

"My posts get no reach" and "I get no interviews" are *different systems*. Fixing the feed will never get someone hired. `diagnose.mjs` routes a symptom to the surface that owns it, so effort lands in the right place:

node diagnose.mjs --list
node diagnose.mjs --symptom "recruiters can't find me"

The one thing to understand first

**LinkedIn's feed is not open source.** Unlike X, there is no repository of weights to read. What LinkedIn publishes is *architecture and mechanism*: which objectives exist, how they're grouped, what the models consume, what the pipeline stages are.

So this skill sorts every claim into three tiers and keeps them apart:

Tier Example
Documented "a comment is much more valuable than a click"; P(skip) reduces the score; the item prompt contains your headline
Inference from mechanism A link ends feed dwell, therefore trades dwell for a click — stated as inference, not