instagram-superpower
Skill4 starsChanged 4 months agoMITView on GitHub
Full toolkit for Instagram: account analytics via Loore (MCP + REST, https://docs.loore.ai) + downloading reels/posts via Cobalt API (or Loore ig_download). Use when: (1) download a reel/video/photo from Instagram, (2) find top-performing reels of an account, (3) analyze a competitor (views, likes, comments, engagement), (4) manage a watchlist of accounts for monitoring, (5) download + transcribe video, (6) hashtag/location search. Triggers: "download reel", "download from instagram", "analyze account", "viral reels", "top reels", "watchlist", "track account", "competitor instagram", "engagement", links to instagram.com/reel/, instagram.com/p/, instagram.com/stories/. NOT for: YouTube, Twitter, mass spam, auto-likes, auto-follows.
- Reads credentials
- Sends data out
In this repository43 other files it gives its agents.
- agent-browserSkill
- agent-introspectionSkill
- agentos-contentSkill
- carousel-instagramSkill
- chat-archiveSkill
- codex-imageSkill
- content-engineSkill
- cross-reviewSkill
- datawrapperSkill
- dev-pipelineSkill
- excalidrawSkill
- fast-loop-codingSkill
- groq-voiceSkill
- higgsfield-generateSkill
- higgsfield-marketplace-cardsSkill
- higgsfield-product-photoshootSkill
- higgsfield-soul-idSkill
- learningsSkill
- loop-codingSkill
- markdown-newSkill
- mcp-api-buildSkill
- mcp-builderSkill
- memory-auditSkill
- miro-boardSkill
- perplexity-researchSkill
- presentSkill
- reel-radarSkill
- reels-analytics-for-brokersSkill
- reelsSkill
- remote-agent-desktopSkill
- security-auditSkill
- senior-brainstormSkill
- seo-tuneSkill
- server-doctorSkill
- telegram-chipSkill
- threads-contentSkill
- topic-monitorSkill
- transcriptSkill
- twitterSkill
- workshopSkill
- youtube-producerSkill
- youtube-thumbnailSkill
- yt-researchSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 108
Every REST example passes the key header through stdin (-H @- with <<<), never as a -H "Authorization: ..." argument: command arguments are …
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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
Instagram SuperPower
Tools
| Tool | Purpose | Cost |
| Loore | Analytics: profiles, reels, posts, comments, followers/following, hashtags, places, search, reel transcripts, file downloads | Credits per request, see https://loore.ai |
| Cobalt | Download video/photo from Instagram CDN | Free (self-hosted) |
| Groq Whisper | Transcription of downloaded videos | Free |
Documentation and links
| Resource | URL |
| Loore docs | https://docs.loore.ai |
| Loore tool list | https://docs.loore.ai/mcp/tools |
| Loore OpenAPI contract | https://app.loore.ai/api/openapi.json |
| Loore dashboard (API key) | https://app.loore.ai → Настройки → API и MCP |
| Cobalt GitHub | https://github.com/imputnet/cobalt |
Path mapping
Agent machine
| What | Path |
| Download script (Cobalt) | ~/.claude/skills/instagram-superpower/scripts/download.sh |
| Analytics script (Loore REST) | ~/.claude/skills/instagram-superpower/scripts/analyze.sh |
| Watchlist script | ~/.claude/skills/instagram-superpower/scripts/watchlist.sh |
| Cookie deploy script | ~/.claude/skills/instagram-superpower/scripts/deploy-cookies.sh |
| Cookie check script | ~/.claude/skills/instagram-superpower/scripts/check-cookies.sh |
| Loore key | env LOORE_API_KEY (or your agent's secret store) |
| Instagram credentials (for Cobalt) | ~/.secrets/cobalt/accounts.json |
| Instagram cookies (for Cobalt) | ~/.secrets/cobalt/cookies.json |
| Cobalt API key | ~/.secrets/cobalt/api-key |
| Watchlist | ~/.secrets/cobalt/watchlist.json |
| Downloaded files | /tmp/downloads/ |
Cobalt server (self-hosted)
| What | Path |
| Cobalt Docker Compose | /opt/cobalt/docker-compose.yml |
| Cobalt cookies (working copy) | /opt/cobalt/cookies.json |
| Cobalt API keys | /opt/cobalt/keys/api-keys.json |
| Docker container | cobalt-api |
| Cobalt API endpoint | http://127.0.0.1:9000/ (localhost only) |
How it all connects
Loore (cloud) Cobalt server (self-hosted)
───────────────── ────────────────────────────
mcp.loore.ai ◀── MCP / REST ── Your machine
analytics, data Cobalt API (localhost:9000)
credits per request media download (free)
Your machine:
agent → Loore MCP tools (ig_profile, ig_user_reels, ...)
analyze.sh → Loore REST (directly)
download.sh → SSH → Cobalt server → Instagram CDN
watchlist.sh → analyze.sh + download.shLoore — Authentication
One key, two doors. Create the key at https://app.loore.ai → Настройки → API и MCP
(the plaintext is shown once). Keep it in LOORE_API_KEY, never in chat or git.
MCP (preferred for the agent) — add the server to your client config:
{
"mcpServers": {
"loore": {
"type": "http",
"url": "https://mcp.loore.ai/mcp",
"headers": { "Authorization": "Bearer <YOUR_KEY>" }
}
}
}REST (scripts) — the same operation, MCP tool ig_profile = POST /v1/ig-profile:
# The key goes in through stdin (-H @-), so it never shows up in argv / ps.
curl -s -X POST https://mcp.loore.ai/v1/ig-profile \
-H @- -H "Content-Type: application/json" \
-d "{\"username\":\"<username>\",\"spend_authorization_id\":\"$LOORE_SPEND_AUTHORIZATION_ID\"}" \
<<< "Authorization: Bearer $LOORE_API_KEY"Every REST example passes the key header through stdin (-H @- with <<<), never as a
-H "Authorization: ..." argument: command arguments are visible in ps to other users.
Answer envelope: {"status":"ok","result":{...}} or {"status":"error","error":"...","details":{...}}.
Spend approval. Paid calls need the user's approval. State the total cap in
credits once, and after the user's «yes» create a flow budget with
spend_authorization (action: "create", flow_id, approved_cap_credits,
user_approved: true, valid 24 hours); pass the returned spend_authorization_id
in every covered paid call (analyze.sh reads it from LOORE_SPEND_AUTHORIZATION_ID).
Without it Loore answers 428 needs_user_confirmation.
Balance: get_credit_balance (POST /v1/billing-balance) — free, shows creditsRemaining.
Loore — Main tools
MCP tool name ↔ REST path: ig_user_reels ↔ POST /v1/ig-user-reels. Full list:
https://docs.loore.ai/mcp/tools.
Users
| Tool | Description |
ig_profile | Profile by username or user_id: id, followers, following, posts, bio, category |
ig_user_reels | Account reels with views/likes/comments/saves/shares, up to 50 per page |
ig_user_posts | Feed posts (photos, carousels, videos), up to 50 per page |
ig_user_stories | Currently active stories (story id for ig_download) |
ig_followers | Followers (paginated, or collect into a file) |
ig_following | Following (paginated, or collect into a file) |
ig_highlights_list / ig_highlight_items | Highlights and their items |
ig_analytics | Ready metrics for a handle: engagement, content mix, posting windows, hashtags, top reels (depth: quick / full) |
Media (posts/reels)
| Tool | Description |
ig_media | One post/reel by link or shortcode, whole caption |
ig_resolve | What a link is (reel, post, story, profile) and its canonical id |
ig_media_comments / ig_comment_replies | Comments and threads |
analyze_reel | Reel transcript with timecodes, metrics, outlier score vs the author's median, frames on depth: deep |
video_transcribe | Word-timed transcript of a reel |
ig_download | The file of a post/reel/story: downloadUrl (no key, ~15 minutes) |
Search
| Tool | Description |
ig_search | kind: accounts / reels / hashtags / places, plus query |
ig_hashtag_media | Publications under a hashtag, rank: top / recent / reels |
ig_location_media | Publications at a place (location_id from ig_search kind=places) |
Response data
Each media row (ig_user_reels, ig_user_posts, ig_search, ig_hashtag_media) contains:
views— plays (reels and videos only)likes— likes (absent andlikesHidden: truewhen the author hid them)comments— commentssaves,shares,reposts— reels (absent means not received, not zero)takenAt— publish time, ISO-8601 UTCcode— shortcode,url— permalinkcaption— caption text wheninclude_caption: true(500 chars;ig_mediagives the whole caption)author— author handleisPinned,isTrial— leave such rows out of the author's median
Quick start
Account profile
MCP: ig_profile with {"username": "<username>"}. REST:
curl -s -X POST https://mcp.loore.ai/v1/ig-profile \
-H @- -H "Content-Type: application/json" \
-d "{\"username\":\"<username>\",\"spend_authorization_id\":\"$LOORE_SPEND_AUTHORIZATION_ID\"}" \
<<< "Authorization: Bearer $LOORE_API_KEY" | python3 -m json.toolAccount reels (top performing)
MCP: ig_user_reels with {"username": "<username>", "limit": 50}, sort by views. REST:
curl -s -X POST https://mcp.loore.ai/v1/ig-user-reels \
-H @- -H "Content-Type: application/json" \
-d "{\"username\":\"<username>\",\"limit\":50,\"spend_authorization_id\":\"$LOORE_SPEND_AUTHORIZATION_ID\"}" \
<<< "Authorization: Bearer $LOORE_API_KEY" \
| python3 -c "
import json, sys
items = json.load(sys.stdin).get('result', {}).get('items', [])
for m in sorted(items, key=lambda x: x.get('views', 0), reverse=True)[:5]:
print(f'{m.get(\"views\",0):>10,} views | {m.get(\"likes\",0):>6,} likes | {m[\"url\"]}')
"Download reel (via Cobalt)
bash ~/.claude/skills/instagram-superpower/scripts/download.sh "https://www.instagram.com/reel/<CODE>/"Download + transcribe
bash ~/.claude/skills/instagram-superpower/scripts/download.sh "https://www.instagram.com/reel/<CODE>/" transcribeCheck Loore balance
MCP: get_credit_balance (free). REST:
curl -s -X POST https://mcp.loore.ai/v1/billing-balance \
-H @- -H "Content-Type: application/json" -d '{}' \
<<< "Authorization: Bearer $LOORE_API_KEY"Typical scenarios
1. "Show top reels of a competitor"
bash scripts/analyze.sh <username> 5 142. "Download this reel and summarize it"
bash scripts/download.sh "<url>" transcribeCobalt downloads → Groq transcribes → agent analyzes.
Alternative without Cobalt and Groq: Loore analyze_reel (depth: "quick" — caption,
metrics, transcript with timecodes) or video_transcribe (word-timed transcript).
3. "Add to monitoring"
bash scripts/watchlist.sh add <username>
bash scripts/watchlist.sh scan 3 144. "Full competitor breakdown"
analyze.sh <username> 5 14→ top-5 reels + metricsdownload.sh <url> transcribefor top-3 (or Looreanalyze_reelper reel)- Agent analyzes: hook → pain → solution → CTA
5. "What's viral in a hashtag"
MCP: ig_hashtag_media with {"hashtag": "aitools", "rank": "top"} (or "reels" for video only).
REST:
curl -s -X POST https://mcp.loore.ai/v1/ig-hashtag-media \
-H @- -H "Content-Type: application/json" \
-d "{\"hashtag\":\"aitools\",\"rank\":\"top\",\"limit\":50,\"spend_authorization_id\":\"$LOORE_SPEND_AUTHORIZATION_ID\"}" \
<<< "Authorization: Bearer $LOORE_API_KEY"Places: ig_search {"kind": "places", "query": "..."} → ig_location_media with the place id.
Pagination
List tools answer hasMore and, when it is true, nextCursor. Pass it back unchanged
as cursor for the next page (limit up to 50; above 50 is refused, not clamped):
curl -s -X POST https://mcp.loore.ai/v1/ig-user-reels \
-H @- -H "Content-Type: application/json" \
-d "{\"username\":\"<username>\",\"limit\":50,\"cursor\":\"<nextCursor>\",\"spend_authorization_id\":\"$LOORE_SPEND_AUTHORIZATION_ID\"}" \
<<< "Authorization: Bearer $LOORE_API_KEY"Send the same spend_authorization_id with every page: a page without it is a new
unapproved spend and gets 428 needs_user_confirmation.
A cursor is valid for 6 hours; after that the walk restarts from page one.
Rate Limits
Loore
- Budgets per key, per minute and per day, depending on the plan; current values in the dashboard
- On
429— wait (Retry-After/details.retry_after_seconds) and retry; if the wait is longer than a couple of minutes (details.scope: "day"), stop and tell the user when to rerun - On
403 private_account— skip that account; on401/402/ other403/428— stop and tell the user - Heavy calls (
analyze_reel,ig_download) cost more — do not loop them "just in case"
Cobalt (downloads)
- Rate limits Instagram: 20 downloads/hour per 1 account
- Pause between requests: minimum 3 seconds
- Batch downloads (>5): pause 10-15 seconds
- Strictly sequential, never parallel
Optimal strategy
- Analytics (Loore): one meaningful call instead of ten;
ig_analyticsgives a whole account in one request - Downloads (Cobalt): 20/hour, sequential
- Loore can also download (
ig_download, paid per file) — use it when Cobalt is not set up
Maintenance
Loore
curl -s -X POST https://mcp.loore.ai/v1/billing-balance \
-H @- -H "Content-Type: application/json" -d '{}' \
<<< "Authorization: Bearer $LOORE_API_KEY"
# Key in dashboard: https://app.loore.ai → Настройки → API и MCP (revoke a leaked key there)Cobalt
ssh root@<your-server-ip> "docker ps --filter name=cobalt-api --format '{{.Status}}'"
ssh root@<your-server-ip> "docker logs cobalt-api --tail 20"
ssh root@<your-server-ip> "cd /opt/cobalt && docker compose restart cobalt-api"Updating Instagram cookies (for Cobalt)
- Credentials:
~/.secrets/cobalt/accounts.json - Login via browser automation (agent-browser via CDP)
- Login using
Input.dispatchKeyEvent(character by character) - Button:
div[role="button"] - 2FA: TOTP →
input[name="verificationCode"] - Cookies:
Network.getCookies→.instagram.com→cookies.json - Deploy:
bash scripts/deploy-cookies.sh
Security
- Loore key: env
LOORE_API_KEYonly, headerAuthorization: Bearer(never in URL, chat or git) - Cobalt: ONLY
127.0.0.1:9000on the Cobalt server - Cookies/credentials: ONLY
~/.secrets/cobalt/(chmod 600) - Forbidden: logging keys/cookies, printing in chat, committing to git
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.

