Security audit
Security checks for vulnerabilities and agentic risk
This skill is a straightforward guide for using a third-party public social-media scraping API, with privacy and retention caveats but no hidden or destructive behavior found.
Before installing, understand that searches, profile handles, post URLs, content IDs, filters, and similar lookup terms will be sent to ScrapeCreators using your API key. Use it for public, non-sensitive research, limit pagination and comment/transcript collection to what you actually need, and be deliberate about saving JSON or CSV outputs because those files may retain scraped personal or business data locally.
The skill routes user-supplied handles, search terms, URLs, and similar parameters to a third-party scraping API, but it does not clearly warn that these inputs are being transmitted off-platform. This creates a privacy and consent risk, especially if users provide sensitive targets, investigative queries, or proprietary search terms assuming processing is local.
This example sends a requested social-media handle to an external third-party service using the configured API key. In this skill context, external transmission is expected, but it remains security-relevant because user inputs and lookups leave the local trust boundary and may expose sensitive investigative intent or personal data.
curl -s -G "https://api.scrapecreators.com/v1/tiktok/profile" \
--data-urlencode "handle=khaby.lame" \
-H "x-api-key: $SCRAPECREATORS_API_KEY" | jq .
This request transmits a target handle to an external scraping service to obtain platform data. Although core to the skill’s purpose, the transmission still poses privacy and operational risks because it reveals the subject of the lookup and depends on a non-local service.
# First page
curl -s -G "https://api.scrapecreators.com/v3/tiktok/profile/videos" \
--data-urlencode "handle=khaby.lame" \
-H "x-api-key: $SCRAPECREATORS_API_KEY" | jq .
The pagination example continues sending lookup context and cursors to the external provider, potentially broadening the amount of retrieved and exposed data. Repeated paginated calls can amplify privacy impact and data over-collection if the agent keeps fetching beyond the user’s actual need.
-H "x-api-key: $SCRAPECREATORS_API_KEY" | jq .
# Next page (using max_cursor from previous response)
curl -s -G "https://api.scrapecreators.com/v3/tiktok/profile/videos" \
--data-urlencode "handle=khaby.lame" \
--data-urlencode "max_cursor=CURSOR_VALUE" \
-H "x-api-key: $SCRAPECREATORS_API_KEY" | jq .
The generic request pattern explicitly encourages sending user-provided parameters such as handles, search queries, and URLs to a third-party API. While it properly recommends safe URL encoding, it does not address the separate risk that these potentially sensitive inputs are transmitted externally.
Every call follows the same shape. Always use -G with --data-urlencode to safely pass query parameters — this prevents shell injection from user-provided values (handles, search queries, URLs) and properly encodes special characters:
curl -s -G "https://api.scrapecreators.com/v1/{platform}/{endpoint}" \
--data-urlencode "param=value" \
-H "x-api-key: $SCRAPECREATORS_API_KEY"
This chaining example starts by sending a profile lookup to the external API, establishing a pattern of multi-step third-party enrichment. Multi-step workflows increase the volume of externalized user intent and can compound exposure when the subject matter is sensitive.
# 1. Get profile
PROFILE=$(curl -s -G "https://api.scrapecreators.com/v1/tiktok/profile" \
--data-urlencode "handle=creator_name" \
-H "x-api-key: $SCRAPECREATORS_API_KEY")
The second chained request expands the external transmission from identity lookup to content retrieval, increasing the amount of collected subject data. In security terms, the risk is not code injection here but unnecessary third-party disclosure and over-collection.
-H "x-api-key: $SCRAPECREATORS_API_KEY")
# 2. Get their recent posts
POSTS=$(curl -s -G "https://api.scrapecreators.com/v3/tiktok/profile/videos" \
--data-urlencode "handle=creator_name" \
-H "x-api-key: $SCRAPECREATORS_API_KEY")
This step transmits a derived video identifier to the external API to obtain comments, deepening subject profiling and potentially collecting large amounts of user-generated content. The chaining makes the workflow more privacy-invasive than a single lookup, especially if used without clear user understanding.
VIDEO_ID=$(echo "$POSTS" | jq -r '.aweme_list[0].aweme_id')
COMMENTS=$(curl -s -G "https://api.scrapecreators.com/v1/tiktok/video/comments"
--data-urlencode "video_id=$VIDEO_ID"
-H "x-api-key: $SCRAPECREATORS_API_KEY")
The skill’s manifest says it should only be used for retrieving data directly from social platforms, but the documentation expands into saving outputs locally and performing downstream analytics. That scope creep can cause the agent to apply the skill in contexts it was explicitly supposed to avoid, increasing the chance of unintended local data handling and policy bypass.
This example both transmits a handle to a third-party service and saves the resulting data locally. That combination expands risk from external disclosure to local persistence, creating retention and secondary-use concerns inconsistent with the skill’s narrow retrieval-only positioning.
# Save raw JSON
curl -s -G "https://api.scrapecreators.com/v1/instagram/profile" \
--data-urlencode "handle=natgeo" \
-H "x-api-key: $SCRAPECREATORS_API_KEY" > natgeo_profile.json
This example fetches external data and converts it into a local CSV artifact, making subsequent reuse and redistribution easier. The security concern is not the CSV transform itself but that the skill encourages persistence and analysis workflows beyond simple retrieval while using third-party scraped data.
-H "x-api-key: $SCRAPECREATORS_API_KEY" > natgeo_profile.json
# Save as CSV (extract specific fields with jq)
curl -s -G "https://api.scrapecreators.com/v3/tiktok/profile/videos" \
--data-urlencode "handle=creator" \
-H "x-api-key: $SCRAPECREATORS_API_KEY" \
| jq -r '.aweme_list[] | [.aweme_id, .desc, .statistics.play_count, .statistics.digg_count] | @csv' \
This metric-extraction example still relies on transmitting the user’s target to an external service, then performs local analysis on the returned data. In context, it broadens the skill from simple fetches into analytics, increasing the chance of overuse and unintended processing of third-party data.
# Get engagement stats from a profile
curl -s -G "https://api.scrapecreators.com/v1/tiktok/profile" \
--data-urlencode "handle=creator" \
-H "x-api-key: $SCRAPECREATORS_API_KEY" \
| jq '{followers: .statsV2.followerCount, following: .statsV2.followingCount, likes: .statsV2.heartCount}'
This aggregation workflow can encourage bulk retrieval and post-processing of subject data after transmission to an external service. The combination of external collection and analytics raises privacy and governance concerns beyond a narrow one-off lookup.
# Average engagement across recent posts
curl -s -G "https://api.scrapecreators.com/v3/tiktok/profile/videos" \
--data-urlencode "handle=creator" \
-H "x-api-key: $SCRAPECREATORS_API_KEY" \
| jq '[.aweme_list[] | .statistics.digg_count] | (add / length)'
Fetching transcripts sends content identifiers to the third-party API and retrieves potentially rich textual data for further analysis. Transcript data can be more sensitive than simple metadata because it exposes full content context and is easier to search, summarize, or repurpose.
# Fetch transcript of a TikTok video
curl -s -G "https://api.scrapecreators.com/v1/tiktok/video/transcript" \
--data-urlencode "video_id=VIDEO_ID" \
-H "x-api-key: $SCRAPECREATORS_API_KEY" | jq -r '.data.transcript'
This search/filter example transmits targeting criteria like follower thresholds and audience country to a third-party service, which may reveal business strategy, investigative focus, or profiling intent. The risk is elevated because searches can encode more sensitive intent than simple public-profile retrieval.
# Find TikTok creators by niche with audience filters
curl -s -G "https://api.scrapecreators.com/v1/tiktok/creators/popular" \
--data-urlencode "min_followers=100000" \
--data-urlencode "audience_country=US" \
-H "x-api-key: $SCRAPECREATORS_API_KEY" | jq '.data'
Even the error-handling example performs an external request with a user-provided handle, meaning debugging steps still transmit data off-platform. This matters because users may not realize that troubleshooting requests have the same privacy implications as normal operation.
# Check status code
curl -s -G -o response.json -w "%{http_code}" \
"https://api.scrapecreators.com/v1/tiktok/profile" \
--data-urlencode "handle=creator" \
-H "x-api-key: $SCRAPECREATORS_API_KEY"
The manifest excludes local file analysis, yet the skill later instructs saving scraped responses to local JSON and CSV files for further use. This inconsistency can mislead an agent into handling local artifacts under a skill that was meant only for remote retrieval, weakening safety boundaries.
No suspicious patterns detected.