# creatorloop (pypi · creatorloop)

Instagram content studio for Claude: analytics, competitors, Reels, publishing queue.

- Trust score: 89/100 (high trust)
- Change this week: +3
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-11-05

## Components

- pypi · `creatorloop`: 89/100 (this document), [markdown](https://verifymcp.io/servers/pandich93-creatorloop/creatorloop.md), [page](https://verifymcp.io/servers/pandich93-creatorloop/creatorloop)

## Channel facts

- Registry: `pypi`
- Package: `creatorloop`
- Version: `0.3.0`
- Transport: `stdio`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog)

Scored 2026-11-05.

- **Supply Chain Security**: 100/100
  - No malware found by supply-chain analysis.
  - No known CVEs affecting this package version or its production dependencies.
  - Runs hatchling.build at install time, a recognised build step with no custom scripting around it.
  - 1 of 34 dependencies flagged as unhealthy.
- **Provenance & Transparency**: 100/100
  - Source repository is publicly reachable at the declared URL.
  - Cryptographically verified build provenance (signed, bound to pandich93/creatorloop).
  - Clear OSI-approved license (MIT).
  - Actively maintained (last published 17 days ago).
  - Publishes a security disclosure policy (SECURITY.md).
- **Schema Quality & AI Usability**: 80/100
  - 100% of prompts and resources have a non-trivial description (not blank, and not just the item's name).
  - AI-judged instruction clarity (good).
  - Tool/resource definitions use about 3015 tokens (~88/item across 34 items; 34 tools + 0 resources), lean.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 57/100
  - Stability observed for 17 of 30 days with no destabilising changes; credit accrues until the full window elapses.
- **Tool Coverage**: 85/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 49% of tool parameters carry a description.
  - Structured output schemas are declared (97% of tools); any adoption earns full credit.
- **Tool Safety**: 100/100
  - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.
  - All 4 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation.
  - An AI judge read all 35 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 100/100
  - Implements a current MCP spec version (2026-07-28).

## Install

### How do I install the creatorloop MCP server?

creatorloop runs locally as a PyPI package, launched with uvx creatorloop. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

### Claude

```bash
claude mcp add pandich93-creatorloop -- uvx creatorloop
```

### Cursor

```json
{
  "mcpServers": {
    "pandich93-creatorloop": {
      "command": "uvx",
      "args": [
        "creatorloop"
      ]
    }
  }
}
```

### VS Code

```json
{
  "servers": {
    "pandich93-creatorloop": {
      "command": "uvx",
      "args": [
        "creatorloop"
      ]
    }
  }
}
```

### Codex

```bash
codex mcp add pandich93-creatorloop -- uvx creatorloop
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "pandich93-creatorloop": {
      "type": "local",
      "command": [
        "uvx",
        "creatorloop"
      ],
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add pandich93-creatorloop --command uvx --arg creatorloop
```

### Hermes

```yaml
mcp_servers:
  pandich93-creatorloop:
    command: "uvx"
    args: ["creatorloop"]
```

### Netclaw

```json
{
  "McpServers": {
    "pandich93-creatorloop": {
      "Transport": "stdio",
      "Command": "uvx",
      "Arguments": [
        "creatorloop"
      ]
    }
  }
}
```

### Vellum

```bash
assistant mcp add pandich93-creatorloop -t stdio -c uvx -a creatorloop
```

### Other

```json
{
  "mcpServers": {
    "pandich93-creatorloop": {
      "command": "uvx",
      "args": [
        "creatorloop"
      ]
    }
  }
}
```

## Changelog

Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims.

### 2026-11-05 (score 89, +15)

- [security improvement] Malware scan: unverified → pass

### 2026-11-04 (score 74, +1)

No change was recorded against any check on this day. Stability & Change Management went from 50 to 53. That category is still filling its 30-day observation window: 15 days of observed history at the previous scan, 16 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-11-03 (score 73, −15)

- [security regression] Malware scan: pass → unverified

### 2026-11-02 (score 88, +16)

- [security improvement] Malware scan: unverified → pass

### 2026-10-31 (score 72, −14)

- [security regression] Malware scan: pass → unverified

### 2026-10-29 (score 86, +1)

No change was recorded against any check on this day. Stability & Change Management went from 30 to 33. That category is still filling its 30-day observation window: 9 days of observed history at the previous scan, 10 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-10-27 (score 85, +1)

No change was recorded against any check on this day. Stability & Change Management went from 23 to 27. That category is still filling its 30-day observation window: 7 days of observed history at the previous scan, 8 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-10-24 (score 84, +1)

No change was recorded against any check on this day. Stability & Change Management went from 13 to 17. That category is still filling its 30-day observation window: 4 days of observed history at the previous scan, 5 at this one. The score rises as the window fills, whether or not the server changes.

## MCP tools (34)

### `get_my_profile` (~37 tokens)

Own profile: username, bio, followers, following, posts count. Also saves a daily follower snapshot.

Input parameters:

- `refresh` (boolean)

### `get_my_media` (~47 tokens)

Latest own posts/Reels (newest first) with public counters: likes, comments, type, permalink, timestamp.

Input parameters:

- `limit` (integer)
- `refresh` (boolean)

Output parameters:

- `result` (array)

### `get_media_insights` (~61 tokens)

Private insights for ONE own post/Reel/Story: reach, views, saved, shares, profile visits, follows,
Reels average watch time (ms). Metrics the API refuses are listed in `unavailable`.

Input parameters:

- `media_id` (string, required)

### `get_my_content_report` (~119 tokens)

Full content report for own account: best/worst posts with hooks, format comparison (Reels vs carousel
vs image), weighted engagement rate, follows/profile visits per 1k reach, Reels watch time,
posting frequency, best weekday/hour, most used hashtags, posts still growing (<72h).
Makes 1 insights request per post (cached), so large max_posts is slower.

Input parameters:

- `days` (integer): Analyse posts published in the last N days
- `max_posts` (integer)
- `rank_by` (string)

### `get_caption_insights` (~86 tokens)

What in captions works: language (uz/ru/en), question in the hook, call-to-action, caption length;
best and worst hooks (first line). Own account is scored by reach, competitors by likes+comments.

Input parameters:

- `days` (integer)
- `max_posts` (integer)
- `username`: Empty = own account (by reach); else a competitor

### `get_account_insights` (~71 tokens)

Account-level totals for the period: reach, views, accounts engaged, interactions, likes, comments,
saves, shares, link taps, follows/unfollows, plus daily reach. Meta data can lag up to 48h.

Input parameters:

- `days` (integer): Period length in days, ending yesterday

### `get_audience_demographics` (~64 tokens)

Audience demographics (top 45 rows). Requires 100+ followers (or 100+ engagements for engaged=True).

Input parameters:

- `breakdown` (string)
- `engaged` (boolean): True = engaged audience, False = all followers
- `timeframe` (string)

### `get_competitor` (~93 tokens)

Public analysis of another PROFESSIONAL (business/creator) account via Business Discovery:
profile, ER by followers, formats, posting frequency, best times, hashtags, top/worst posts.
Saves a daily follower snapshot so growth can be tracked with get_follower_history.

Input parameters:

- `media_limit` (integer)
- `refresh` (boolean)
- `username` (string, required): Instagram username, with or without @

### `compare_accounts` (~75 tokens)

Side-by-side comparison of up to 10 professional accounts (optionally including own account, measured
the same public way): followers, posts/week, avg likes/comments/Reels views, ER by followers, main format.

Input parameters:

- `include_me` (boolean)
- `media_limit` (integer)
- `usernames` (array, required)

### `get_follower_history` (~58 tokens)

Follower growth from local daily snapshots (own account or any competitor fetched before).
History starts from the first time the account was fetched through this server.

Input parameters:

- `days` (integer)
- `username`: Username; empty = own account

### `track_competitor` (~71 tokens)

Add an account to the tracked list (validated with one Business Discovery request).
Tracked accounts are refreshed together by update_tracked_competitors.

Input parameters:

- `note`: Why you track it, e.g. 'AI-blog, uz'
- `username` (string, required): Professional account username, with or without @

### `untrack_competitor` (~33 tokens)

Remove an account from the tracked list. Its history stays in the database.

Input parameters:

- `username` (string, required)

### `list_tracked_competitors` (~22 tokens)

Tracked accounts with the date added and note.

### `update_tracked_competitors` (~83 tokens)

Refresh all tracked accounts (fresh API data, 1-2 requests each): follower change over the stored history,
new posts since the previous update and viral posts (score >= multiplier × the account's median).
Run it regularly (e.g. weekly) to build history.

Input parameters:

- `media_limit` (integer)
- `viral_multiplier` (number)

### `find_viral_posts` (~85 tokens)

Posts that outperformed their account: Reels by views, other formats by likes+comments,
compared with the account's own median. Good source of proven topics and hooks.

Input parameters:

- `days` (integer): Only posts from the last N days
- `media_limit` (integer)
- `multiplier` (number)
- `username`: Empty = all tracked accounts

### `get_competitor_post` (~203 tokens)

Analyse ONE post of a professional account by its URL, without a browser: caption, hook, likes, comments,
Reels views, format, how many times better than the account's median, and all slides downloaded to
\<content folder>/_refs/<username>_<code>/ (videos: cover image). With show_slides the slide previews are returned as
images (up to 20, 400 px wide) so their text and design can be read directly. Pass username: the author can't be
detected from the URL without Meta App Review. Comment texts are not available via the API.

Input parameters:

- `show_slides` (boolean): Return small previews of the slides as images
- `url` (string, required): Instagram post/reel URL, e.g. https://www.instagram.com/p/DeCHjDpjuI8/
- `username`: Author's username (required in practice: oEmbed auto-detection needs Meta App Review)

### `get_post_history` (~61 tokens)

Growth curve of one post (own or competitor) from stored snapshots: points by post age,
gain per day and status growing/plateau. Snapshots are added on every fresh fetch (max 1 per hour).

Input parameters:

- `media_id` (string, required)

### `search_hashtag` (~97 tokens)

Top (most popular) or recent (last 24h) public posts for a hashtag, with a trend summary:
formats, avg likes/comments, related hashtags, caption length, posting times.
Author usernames are NOT available from the API. Uses 1 of 30 weekly hashtag slots for a new tag.

Input parameters:

- `hashtag` (string, required): Hashtag with or without #
- `limit` (integer)
- `mode` (string)

### `compare_hashtags` (~60 tokens)

Compare several hashtags by their top posts: avg likes/comments, dominant format, Reels share.
Each NEW hashtag uses 1 of 30 weekly slots; check get_api_status first.

Input parameters:

- `hashtags` (array, required)
- `limit` (integer)

### `plan_hashtag_research` (~66 tokens)

Dry run before search_hashtag / compare_hashtags: which tags are free (queried within the last 7 days),
which need a new slot out of 30 per week, and whether they all fit. Makes no API calls.

Input parameters:

- `hashtags` (array, required)

### `prepare_post` (~263 tokens)

Step 1 of publishing (does NOT publish). Starts a BACKGROUND job and returns job_id immediately.
Folder with ONE .mp4/.mov -> Reel: validates the video (3 s-15 min, <=300 MB, moov at front), uploads it straight
to Meta (resumable upload; tunnel only as fallback), cover frame and feed sharing from optional reel.json
{"cover_ms": 1667, "share_to_feed": true}; images in that folder are ignored. Processing takes 1-10 min.
Otherwise 1-10 images (sorted by file name) -> photo/carousel: converts to JPEG, serves them through a temporary
Cloudflare tunnel, creates containers and waits until Meta has processed them (30-120 s).
Caption: caption.txt (text before a '---' line). Poll get_post_job(job_id) every ~15-30 s. When done it returns
creation_id and a preview: show it to the user and call publish_media only after their explicit confirmation.

Input parameters:

- `caption`: Empty = caption.txt from the folder (text before a '---' line)
- `folder` (string, required): Folder in the content folder (e.g. 'my-first-reel') or an absolute path in it

### `get_post_job` (~41 tokens)

Status of a prepare_post job: running (with current stage), done (with creation_id + preview) or failed.

Input parameters:

- `job_id` (string, required)

### `publish_media` (~52 tokens)

Step 2: PUBLISHES the prepared post to Instagram (public, irreversible from here).
Call ONLY after the user explicitly confirmed this exact preview.

Input parameters:

- `creation_id` (string, required): creation_id returned by prepare_post

### `list_publications` (~33 tokens)

Prepared and published posts made through this server, plus the 24h publishing quota.

Input parameters:

- `limit` (integer)

### `queue_post` (~177 tokens)

Schedule a post/carousel/Reel to be PUBLISHED AUTOMATICALLY at a time (default slots Tue/Thu/Sun 19:00 local).
Call first with confirm=false and show the preview (type, caption hook, local time) to the user. Call again with
confirm=true ONLY after the user explicitly approved this post AND this time. The server fingerprints caption and
media: if they change before the slot, the post is blocked. Late by more than 3 h (laptop asleep) = missed.

Input parameters:

- `confirm` (boolean): false = preview only; true = add to the queue (after approval)
- `folder` (string, required): Folder inside the content folder (same as prepare_post)
- `when`: 'next' / empty = next free slot, or 'YYYY-MM-DD HH:MM' local

### `list_queue` (~49 tokens)

Publishing queue: scheduled posts (and with include_done=true also published/failed/missed/blocked/cancelled),
local times, plus the next free default slots.

Input parameters:

- `include_done` (boolean)

### `cancel_queued` (~31 tokens)

Remove a scheduled post from the queue (nothing is deleted on Instagram).

Input parameters:

- `queue_id` (integer, required)

### `post_reviews` (~108 tokens)

Post-mortems for the daily report: own posts (last 7 days) that crossed 24 h or 72 h since the last run,
with fresh insights vs the median of own mature posts of the same format (x_median), and growth 24 h -> 72 h.
Each post is reported once per stage. Explain the numbers to the user and give one concrete takeaway.

Input parameters:

- `mark_reported` (boolean): False = just look, do not mark posts as reported

### `get_new_comments` (~129 tokens)

Comments under own posts that you have not answered yet, including follow-ups inside threads you already
answered (new since the last call by default; they are marked as seen). For each: author, text, post, reply_to
(the top-level comment to answer in that thread). Draft replies in the comment's language and show them to the
user; send with reply_comment(reply_to) only after explicit approval. Needs instagram_manage_comments.

Input parameters:

- `days` (integer): Posts published in the last N days
- `only_new` (boolean): false = all unanswered, including ones shown before

### `reply_comment` (~66 tokens)

PUBLICLY replies to a comment under your post. Call ONLY after the user explicitly approved this exact text.

Input parameters:

- `comment_id` (string, required): reply_to from get_new_comments (the top-level comment of the thread)
- `message` (string, required): Reply text exactly as the user approved it

### `get_api_status` (~44 tokens)

Diagnostics: token validity, days until expiry (warns 14 days ahead), last rate-limit usage headers,
hashtag quota for the rolling 7 days, cache location.

### `open_setup` (~143 tokens)

Connect creatorloop to the user's Instagram account: opens a local setup page in their browser (Meta app,
Facebook login, account choice). Use when settings are missing or the token expired. Never ask for an App Secret
or a token in the chat. Call again to check progress: once it reports done, every tool works without restarting
Claude Desktop. Does not work inside Docker: there the user runs `creatorloop-auth login` (README, Docker).

Input parameters:

- `again` (boolean): true: connect again after a finished setup (expired or revoked token)
- `open_browser` (boolean): false: only return the link, e.g. when the browser is elsewhere

### `clear_cache` (~51 tokens)

Drop cached API responses so next calls fetch fresh data. Follower history is not affected.

Input parameters:

- `prefix`: me: | media: | bd: | hashtag: ; empty = everything. Snapshots are kept.

### `restart_server` (~127 tokens)

Restart this MCP server so updated .py code is loaded, without restarting Claude Desktop.
The process exits ~1 s after replying; Claude Desktop starts it again on the next call. Use only after a code
update (get_api_status shows restart_needed=true). Wait 10-20 s before the next call.
Refuses when dependencies/version changed: then Claude Desktop must be restarted manually.

Input parameters:

- `exit_code` (integer): 1 = report as crash (Desktop restarts it)
- `force` (boolean): Restart even if pyproject.toml/uv.lock changed (will likely fail)

## Diagnostics

Captured diagnostic sections: Provenance, Install scripts, Dependencies. The full working is on the page: https://verifymcp.io/servers/pandich93-creatorloop/creatorloop#diagnostics

## Score history

- 2026-11-05: 89
- 2026-11-04: 74
- 2026-11-03: 73
- 2026-11-02: 88
- 2026-11-01: 72
- 2026-10-31: 72
- 2026-10-30: 86
- 2026-10-29: 86
- 2026-10-28: 85
- 2026-10-27: 85
- 2026-10-26: 84
- 2026-10-25: 84
- 2026-10-24: 84
- 2026-10-23: 83
- 2026-10-22: 83
- 2026-10-21: 82
- 2026-10-20: 82
- 2026-10-19: 81
- 2026-10-18: 50
- 2026-10-17: 50
- 2026-10-16: 50
- 2026-10-15: 50
- 2026-10-14: 50
- 2026-10-13: 50

## Common questions

### What is the creatorloop MCP server?

creatorloop is an MCP server listed in the public MCP registry as io.github.pandich93/creatorloop. Instagram content studio for Claude: analytics, competitors, Reels, publishing queue. This page covers its PyPI package (creatorloop).

### Is the creatorloop MCP server safe to use?

creatorloop scores 89 out of 100 on VerifyMCP. We found no known CVEs affecting it as of 5 November 2026. Its build provenance is signed and verified. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

### What tools does the creatorloop MCP server expose?

creatorloop exposes 34 tools: get_my_profile, get_my_media, get_media_insights, get_my_content_report, get_caption_insights, and 29 more. Their descriptions and schemas cost roughly 2,806 tokens of context every time the server is loaded.

### Is the creatorloop MCP server still maintained?

creatorloop is still listed as active in the MCP registry. We last reached this channel on 5 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

### What licence is the creatorloop MCP server under?

creatorloop declares the MIT licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.

## Links

- PyPI project: https://pypi.org/project/creatorloop/
- Socket report: https://socket.dev/pypi/package/creatorloop
- Repository: https://github.com/pandich93/creatorloop
- Changelog RSS feed: https://verifymcp.io/servers/pandich93-creatorloop/creatorloop.xml
- Changelog JSON feed: https://verifymcp.io/servers/pandich93-creatorloop/creatorloop.json
- HTML version of this page: https://verifymcp.io/servers/pandich93-creatorloop/creatorloop