Security audit
Security checks for vulnerabilities and agentic risk
Postemi’s publishing workflow is coherent, with sensitive actions disclosed and optional billing and add-ons kept under user control.
Grant access only to intended workspaces and channels, and review content and targets before publishing or deleting posts. Selected media is uploaded to Postemi. Optional add-ons can affect multiple local agents; select a specific agent when desired. Subscription payments remain your responsibility.
The supplied commands manage agent skills rather than schedule or publish social posts. Installing and removing skills across local agents is an undeclared capability that changes agent environments, not merely a supporting implementation detail of social publishing.
The supplied commands retrieve analytics rather than schedule or publish posts. Analytics retrieval is a distinct user-facing capability not disclosed in the declared description, rather than an implementation detail of scheduling or publishing.
Billing and subscription management are distinct user-facing capabilities not disclosed by the scheduling and publishing description. Initiating a paid subscription checkout is more than a supporting implementation detail. The chunk does not demonstrate social posting behavior or a completed payment; it documents billing operations and their expected outputs.
The supplied workflow focuses on analytics retrieval and repairing missing content associations, rather than scheduling or publishing posts. Analytics is a distinct user-facing capability not disclosed in the description, and the association repair supports that undeclared capability. No unrelated resource access or malicious behavior is evident.
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
Postemi can generate video, and do not call those routes: they 404.
- **No image or video generation of any kind.** Media comes from files the
user already has, via `postemi upload`.
- **No channel deletion.** `DELETE /integrations/:id` was deliberately removed
from the public API. Disconnecting a channel is a signed-in, admin action in
the web app.
- **Nothing outside the channel table at the top of this file.**
The commands at L1033–L1036 manage locally installed agent skills; L1034 explicitly targets every detected agent, including Claude Code, Codex, Cursor, and Grok Build. Modifying other agents' skill installations is not necessary to schedule or publish social posts and is not declared in the manifest description. The user-consent requirement at L1039–L1043 reduces unintended execution but does not resolve this capability mismatch.
No suspicious patterns detected.