# Jakeschincariol/instagram-agent-skill for Claude Content Planning > Jakeschincariol/instagram-agent-skill packages 13 Claude skills for preparing Instagram Reels, captions, carousels, stories, DMs, and audits. It offers local editorial checks, but it does not publish posts or document a complete live analytics feed. **Jakeschincariol/instagram-agent-skill** — Thirteen Claude skills that run an Instagram account: Reels off 26 scored hook formulas, a viral research skill, a caption linter, and a humanizer. Free, MIT. - Repository: https://github.com/Jakeschincariol/instagram-agent-skill - Stars: 2,441 · Forks: 541 - Language: Python - License: MIT - Published: 2026-10-16 · Updated: 2026-10-16 · Language: en - Canonical page: https://hysenlabs.com/projects/jakeschincariol-instagram-agent-skill ## What the 13 Instagram skills cover Jakeschincariol/instagram-agent-skill packages 13 Instagram-oriented Claude skills under commands including `/ig-reel`, `/ig-caption`, `/ig-profile`, and `/ig-plan`. It is intended for creators and account operators who want help preparing material around a post. The README states that the setup requires no signup or API key and does not connect to Instagram. Its publishing boundary is direct: the skills write and analyze material, while the account owner decides what to post. The collection extends beyond captions. `/ig-reel` turns one idea into a Reel with three hooks selected from 26 formulas, then adds a script, on-screen text, and a timed beat sheet. `/ig-carousel` prepares cover copy, slide copy, and image files. `/ig-story` produces a daily sequence, assigns a purpose to each sticker, and maps a DM funnel. `/ig-profile` applies a 12-part rubric and rewrites profile text in fix-first order. Other commands handle comments, replies, DMs, weekly planning, repurposing, research, humanization, and post audits. `/ig-viral`, for example, is described as finding material in a niche and ranking it by how far each result exceeds the source account's median. This is a content operating procedure built around named Claude modes. It does not claim a guaranteed distribution result or document automatic posting. ## How skill instructions and Python checks cooperate The primary layer consists of `SKILL.md` folders. Claude uses each folder as a mode with task-specific instructions. The README says a skill can be copied into Claude's skill directory, installed as a plugin, placed under a project's `.claude/skills/` directory, or pasted at the top of a chat. The full installation retains five Python tools. The paste-a-`SKILL.md` approach loses those tools, which the README identifies as important to `/ig-reel` and `/ig-human`. The Python tools operate on local text. According to the README, they have no dependencies, make no network requests, and upload nothing. `hookscore.py` ranks candidate hooks and identifies a weak scoring dimension. It can also report a deal-breaker explanation, such as opening with "stop scrolling" or beginning with a greeting. `beats.py` estimates how long each line takes to speak, assigns timecodes, and checks several editing risks. It flags a hook beyond three seconds, an individual beat beyond four seconds, a passage with no concrete content, and a script that never loops back toward its opening. `caption.py` displays the caption area shown before the feed expands the text, then checks total length, the first line, concrete wording, hashtags, calls to action, and requested search terms. These are deterministic editorial checks, not live Instagram analytics. The README does not document a supported Python version range, logging behavior, or a formal output schema. That makes the scripts useful for repeatable review, but their scoring rules remain project-defined judgments rather than guarantees about audience response. ## Installing the skills and configuring the voice file The README offers a direct Claude installation path. A user can paste the repository URL into Claude, ask it to install the skill, and then confirm that `/ig-reel` works. This is the shortest documented route when Claude is already available. For Claude Code, the README gives these commands: ```bash git clone https://github.com/Jakeschincariol/instagram-agent-skill.git cp -r instagram-agent-skill/skills/ig-* ~/.claude/skills/ ``` The second command copies the repository's `ig-*` directories into the user's global Claude skills directory. A project-local setup uses the same skill folders under the current repository's `.claude/skills/` directory, although the README does not provide a separate shell command for that operation. The documented plugin workflow is: ``` /plugin marketplace add Jakeschincariol/instagram-agent-skill /plugin install instagram-agent ``` After installation, the README asks users to prepare `templates/voice.md`. It can be copied to `~/.claude/instagram/voice.md` and completed manually. Alternatively, the user can give Claude three of their own Reels and ask it to write the voice file from those examples. Every skill reads this file. Its documented purpose is to keep generated language and spoken scripts aligned with the account's established voice, but the README does not define a strict schema or explain how conflicts between the voice file and individual prompts are resolved. ## Running a first Reel and caption review A first use can begin with draft text. Place the candidate hooks in `hooks.txt`, one per line, and run: ```bash python3 hookscore.py hooks.txt ``` The README shows a report with a score, a hook, the weakest scoring factor, and a deal-breaker explanation where applicable. Hooks can receive `STRONG`, `OK`, or `WEAK` labels. Those labels come from the project's scoring logic and should not be read as a prediction of reach. After choosing a script, place it in `script.txt` and generate a timed beat sheet: ```bash python3 beats.py script.txt --target 30 ``` The README's sample reports 73 words, about 26.6 seconds at 165 wpm, and a 30-second target. It labels sections as `HOOK`, `MID`, and `CTA`, then identifies long beats, missing concrete details, and whether the ending repeats a phrase from the hook. A new file will produce output based on its own words and selected target rather than reproducing the sample figures. Caption review has its own command: ```bash python3 caption.py caption.txt --keywords "client contracts,freelance pricing" ``` The tool prints a box representing the initial feed display and checks total length, first-line length, concrete language, hashtags, call count, and keyword presence. The README also states that the tool enforces a limit of five hashtags per post rather than thirty. This helps revise a caption before opening Instagram, but it does not submit that caption to Instagram or fetch live display data from the service. ## Where research, planning, and audit claims stop The repository includes workflow commands beyond creation. `/ig-plan` prepares a weekly schedule with proposed posts, formats, timing, and 10 accounts to engage with. It produces a plan, not a scheduler. The README does not identify an Instagram scheduling integration, so the command should not be described as publishing or committing content to a calendar. `/ig-viral` adds a research stage. According to the README, it finds material working in a selected niche, ranks results by their multiple over each account's median, identifies a formula, and writes a swipe file. This depends on the research capability available in Claude. Unlike the five local Python tools, `/ig-viral` is not described as a dependency-free standalone network program, so its external-data process is not documented in comparable detail. `/ig-reply` sorts a post thread into keyword, lead, substance, question, support, and noise categories before drafting responses. `/ig-dm` prepares keyword delivery, an opening message, a collaboration pitch, and two follow-ups. `/ig-comment` selects one of nine comment types based on the target post. `/ig-audit` reviews prior posts and ranks them by outlier multiple and reach rather than views. The audit command has a real documentation gap. The README does not explain how existing post data enters the skill, whether a creator must supply it manually, or whether a connector is required. The repository can standardize analysis of data it receives, but it is not documented as a complete analytics database or live reporting service. ## Humanizer limits, licensing, and maintenance cost The `/ig-human` skill is described as running two Python scripts. According to the README, the humanizer removes em dashes, generic promotional vocabulary, and invisible watermark characters, then scores the draft against five checks. These operations can flag undesirable phrasing, but they do not prove originality or resolve every platform-policy issue. The README's displayed humanizer command is incomplete, so the exact filenames and options should be checked in the repository before use. The repository is licensed under MIT. That license permits use, copying, modification, and distribution under its conditions, including preservation of the applicable copyright and permission notice. `LICENSE` contains the controlling terms and disclaimer. The repository is not archived, and its last push was on 2026-09-13. It has no GitHub releases, and the README does not define a versioned upgrade process. Changes to skills, hook formulas, or Python checks therefore have to be reviewed through repository updates. Editorial maintenance also depends on Instagram's changing interface. The README says Instagram changed its hashtag allowance on 2025-12-18 and that the caption tool now enforces five hashtags per post. A user relying on that check should inspect `caption.py` during upgrades, especially if the surrounding content requirements change. The project offers useful local review, but it does not document a policy-monitoring schedule or an automated mechanism for updating platform rules. ## Conclusion Adopt this repository if you already use Claude and want a repeatable workflow for preparing Instagram content, especially when hook timing and visible-caption checks matter. Do not adopt it as an automated publisher or a standalone analytics platform. Before installing, verify that Claude Code or the Claude skill format fits your workflow, review `templates/voice.md`, and test the local scripts with drafts that do not contain sensitive material. The first practical checkpoint is whether `/ig-reel` and `/ig-human` produce usable text without extensive cleanup. ## FAQ ### What is Jakeschincariol/instagram-agent-skill used for? It provides 13 Claude skills for preparing Instagram Reels, carousels, captions, stories, profiles, comments, replies, DMs, plans, research, and audits. The README says users handle posting themselves. ### Does Jakeschincariol/instagram-agent-skill need an Instagram API key? No. The README says no signup, API key, or connection is required. It does not document an automated posting integration. ### How do I install Jakeschincariol/instagram-agent-skill in Claude Code? Clone the repository with `git` and copy the `ig-*` directories into `~/.claude/skills/`. The README also documents project-local installation and plugin installation. ### Does Jakeschincariol/instagram-agent-skill publish posts automatically? No. The documented workflow prepares and reviews material, while the account owner publishes it. The README does not identify an Instagram publishing connector. ### What does `voice.md` do in Jakeschincariol/instagram-agent-skill? The README says every skill reads the voice file. Users can complete `templates/voice.md` manually or ask Claude to derive it from three of their own Reels. ## Sources - [Issues](https://github.com/Jakeschincariol/instagram-agent-skill/issues) - [Jakeschincariol/instagram-agent-skill on GitHub](https://github.com/Jakeschincariol/instagram-agent-skill) - [License: MIT](https://github.com/Jakeschincariol/instagram-agent-skill/blob/main/LICENSE) - [README](https://github.com/Jakeschincariol/instagram-agent-skill/blob/main/README.md) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/jakeschincariol-instagram-agent-skill