Security audit
Security checks for vulnerabilities and agentic risk
This skill transparently connects to SimplePost to publish and manage social posts; the main risk is accidental public posting or public media if used with broad instructions.
Install this only if you want an agent to publish, schedule, draft, or manage social posts through SimplePost. Before asking it to publish, verify the exact accounts, content, media, schedule, and privacy settings, especially for TikTok and any uploaded media URLs. Keep API keys and platform credentials server-side or in approved secret storage, and use drafts or previews when intent is uncertain.
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
yarn install
cp scheduler/.env.example scheduler/.env
yarn workspace @simple-post/scheduler db:migrate
yarn workspace @simple-post/scheduler dev
The default prompt is a broad action trigger that directly frames the skill as ready to publish content to connected social accounts. Because publishing is an external side-effect with reputational and operational consequences, a generic invocation phrase increases the chance of accidental or insufficiently reviewed posting, especially when paired with implicit invocation.
The documentation states that returned media URLs must be publicly reachable by social platforms, but it does not clearly warn users that uploaded media may become internet-accessible. In a posting/integration skill, this can cause accidental disclosure of private or internal media if operators assume uploads remain private.
The workflow explicitly instructs the assistant to call mutating tools such as create_post, update_scheduled_post, and discard_scheduled_post, including immediate publishing with postingMode: "now", without requiring an explicit confirmation step or user warning before external state is changed. In an AI-agent setting, this increases the chance of unintended publication, edits, or deletion of social content from ambiguous prompts or model misinterpretation.
The TikTok section documents a server-side default of PUBLIC_TO_EVERYONE when privacy is omitted, but it does not require the agent to surface this consequence prominently before creating, scheduling, or publishing a TikTok post. Because audience visibility is privacy-sensitive and social posting is externally visible, an omitted setting could cause users to publish more broadly than intended.
The documentation explicitly advertises 'S3-compatible media upload with public URLs' but does not warn that uploaded media will be publicly accessible and potentially indexable or shareable beyond the intended audience. In a social-posting product, users may upload embargoed, private, or sensitive assets, so omission of the privacy implications can lead to inadvertent disclosure.
This markdown file documents posting content and media to third-party platforms such as X, Telegram, LinkedIn, YouTube, and Facebook, which is a privacy- and integrity-relevant action. The description does not include any warning that content, media, or credentials may be sent to external services, nor any caution to obtain user confirmation before publishing.
This markdown file applies to SQP-2, and it instructs users to send an x-api-key header while later noting that users manage platform credentials themselves. The document does not include any warning that these values are sensitive secrets or caution against exposing them in client-side or untrusted contexts.
Detected: suspicious.exposed_secret_literal
x-api-key: [REDACTED]