Back to skill

Security audit

Brandiz.io

Security checks for vulnerabilities and agentic risk

Overview

The scanners describe a Brandiz business-content skill and rate it benign, mainly because its reported behavior aligns with that purpose, while noting risk from repeated unpinned `npx brandiz` commands; overall confidence is medium.

The verdict is benign, not risk-free. Before installing or using the skill, consider whether you are comfortable running an unpinned CLI that may access the credentials and local files available to it. The scan was partial: two local references could not be matched to bundled artifacts.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (46)

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

The CLI entry point uses an unpinned npm package, allowing npx to download and execute a release that has not been reviewed. A compromised release could access local files and available credentials; this is a CLI supply-chain risk, not an MCP-server version issue.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 27)May include surrounding context.

md
**CLI (OpenClaw, Hermes, scripts and any agent with a shell).** `npx brandiz` (Node.js 20 or newer).

- The owner makes an API key in Brandiz.io under **Connect AI → Make an API key**, and ticks **Let it schedule and post for me** there if the key may post without asking.
- Set it as `BRANDIZ_API_KEY`, or the owner runs `npx brandiz login` once and pastes it.
- Add `--json` to any command for an answer you can read reliably. Exit code `0` is done, `1` didn't work (the reason is on stderr), `2` the command was typed wrong.
- Never print, log or echo the key, and never write it into a project file.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 37)May include surrounding context.

md
ry number, price, date, discount and result comes from the business's facts or proof, exactly as written.** If it isn't there, leave it out or ask the owner to add it in Brandiz.io. Never invent one. Brandiz.io refuses words that break this and says which.
3. **Write in the business's language and voice**, as `business` gives them.
4. **Posts wait for the owner's Approve** unless the owner ticked **Let it schedule and post for me** for this connection. Then a scheduled post goes out at its time without asking. `schedule_post` and `add_own_post` answer whether the post was approved or is waiting; tell the owner which. CLI: `npx brandiz whoami` says what the key may do.
5. **Making uses the package.** `make_poster` uses one post from the business's package the first time; `make_reel` uses one reel when it renders. Check `package` before making several, and say what it will use. Uploads, the owner's own posts and `make_our_version` use nothing.
6. **Never delete posts, connect or disconne

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

The login command executes an unpinned npm package immediately before the owner supplies an API key. A compromised release could capture that key despite the separate instruction not to print or log it.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

The business-reading command can execute an unreviewed npm release because no version is specified. Compromised CLI code could access business information, local files, and credentials available to the process.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Even the permissions-check command invokes an unpinned package with the agent's local privileges. A malicious release could steal the API key or misuse its posting permissions rather than merely report them.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

This repeats the unpinned CLI execution path for reading the business. A compromised npm release could execute arbitrary code with access to the agent's environment; the read-only business operation does not restrict local code privileges.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Retrieving ideas uses the same unpinned npm executable as other commands. If its release is compromised, the process could expose credentials or local data regardless of this operation's read-only purpose.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Listing channels executes an npm package without fixing its version. A compromised release would run with the agent's local privileges and could obtain the API key and connected-channel information.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

The calendar command can resolve to an unreviewed npm release. Compromise of that release could expose local data or credentials even though the requested API action only reads a calendar.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Checking package usage invokes an unpinned executable from npm. A compromised release could run arbitrary code and access the API key; the benign command purpose does not mitigate package execution privileges.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

The results command shares the unpinned CLI supply-chain exposure. A malicious npm release could access business analytics, credentials, and local files rather than merely retrieve results.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

The working-content command executes an npm package without a version constraint. A compromised release could access the agent's files and credentials, independently of the safeguards for treating returned content as data.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Creating a business-specific idea uses an unpinned npm executable. A compromised release could steal credentials or perform other actions allowed by the key, beyond creating the requested idea.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Listing posts invokes the same unpinned npm package. A compromised release could execute arbitrary local code and expose credentials, so a read-only API action does not remove the supply-chain risk.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Reading a post executes an npm package whose version is unspecified. Compromise of the resolved release could expose the API key and local data, not just the selected post.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Status polling invokes the unpinned npm CLI. A compromised release could run with the agent's filesystem and environment access even though the requested action only checks progress.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Adding post text executes a package that is not pinned to a reviewed release. Malicious package code could steal credentials or alter submitted content, in addition to accessing local files.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Uploading media runs an unpinned npm package with access to the selected file and the agent's environment. A compromised release could upload unintended data or steal credentials instead of limiting itself to the authorized media.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Posting owner-provided media invokes an unpinned executable. A compromised release could access local media and credentials or misuse any posting permissions granted to the key.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Editing a post runs an npm package without an exact version. A compromised release could modify unintended content or steal the API key, subject to the key's available permissions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Poster generation uses the unpinned npm CLI. A compromised release could expose credentials or invoke unintended permitted operations, including actions that consume the business's package.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Reel generation executes an npm release that is not fixed by the skill. A compromised release could steal credentials or misuse authorized generation capabilities and consume package resources.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Scheduling invokes an unpinned npm executable in a workflow that may hold publishing permissions. A compromised release could steal the key or schedule unauthorized content within those permissions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx brandiz'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Stopping a post still executes the unpinned npm CLI with local privileges. A compromised release could expose credentials or disrupt other permitted posting operations instead of only stopping the requested post.

Content

No source excerpt is available for this finding.