ig-autopilot
Skill0 starsChanged 30 days agoMITView on GitHub
Run an Instagram account's posting on autopilot through the official Instagram API - connect the account, queue videos and images, approve them into learned time slots, and let a scheduler check, fix, upload and publish them. Use when the user says "post this to Instagram", "schedule my reels", "set up instagram autopilot", "what's queued", "why didn't it post", or wants hands-off posting.
In this repository3 other files it gives its agents.
- ig-analyzeSkill
- ig-captionSkill
- ig-learnSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 44
Only set approval auto when the user explicitly asks for it, and say what it means: every draft with a caption gets scheduled and posted …
- Distinctive: line 46
Only the post itself is automated.
- Very specific: line 48
Never invent numbers, testimonials or results in captions.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
ig-autopilot
Everything runs through one CLI. In a plugin install it is node "${CLAUDE_PLUGIN_ROOT}/cli/igap.mjs";
in a clone it is node <repo>/cli/igap.mjs (or npm link once and call igap). Below, igap means
whichever applies. Data lives in ~/.instagram-autopilot/ (override with IGAP_HOME).
First run
igap status. If the account is not connected, walk the user through the README's "Connect an account" section. The token is something only they can generate; never ask them to paste it into chat. Have them runigap setupin their own terminal, where the input is hidden.- Hosting: Instagram downloads media from a public URL. Default is a public GitHub repo's release
(
igap config set hosting.repo owner/name; uploads run as the activeghaccount, which needs write access). Any other uploader works throughhosting.provider command(see README). igap config set timezone <IANA zone>if the machine's zone is not the audience's.
The loop
igap add <video> --hook <formula-tag> # draft; write the caption into the queue file
igap analyze <id> # specs, timing, what will be fixed (see ig-analyze)
igap lint <id> # caption check + the 125 characters the feed shows
igap approve <id> [--at "Tue 12:00"] # books the next learned slot unless a time is given
igap schedule # once: a timer runs `igap run` every 15 minEach run tick: refreshes the token when due, pulls insights for recent posts, re-learns the
playbook, then posts at most one due item: analyze → auto-fix a copy (silence, black open, loudness,
9:16 frame, codecs) → caption lint → upload → publish → record features + tweaks in the ledger.
Rules you keep
- Approval is the user's.
approval: manual(default) means nothing posts untiligap approve. Only setapproval autowhen the user explicitly asks for it, and say what it means: every draft with a caption gets scheduled and posted without another look. - Only the post itself is automated. Never automate likes, comments, follows or DMs, and never use browser automation against instagram.com. Both break Instagram's terms and get accounts restricted.
- Never invent numbers, testimonials or results in captions. If a caption needs a figure the user
has not given, leave
{{number}}and tell them. - A
blockeditem says why inblocked_reason. Content problems (video can't be posted, caption errors) need a fix andigap unapprove+approve; network failures retry 3 times on their own.
Diagnosing "it didn't post"
igap status, igap list, then the tail of ~/.instagram-autopilot/autopilot.log. Idle reasons
are logged verbatim: daily cap, minimum gap, nothing due, quota used up.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

