Back to skill

Security audit

LedgerMind

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent local memory installer guide, but it tells users or agents to run a changing remote installer directly, which warrants review before use.

Install only if you trust the LedgerMind publisher and can verify the exact release artifact before running it. Prefer downloading a pinned installer, checking its signature or checksum, and reviewing what files, adapters, services, and model endpoints it will configure. Expect the tool to create persistent local memory and agent integrations, and avoid placing plaintext credentials in its config.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Findings (1)

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:68
Finding

Unverified Remote Installer Downloaded and Executed Directly

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 68 and 76–78
Vulnerability Type: Unverified remote code execution through curl | sh
Risk Level: High

Vulnerable Code:

bash
curl -fsSL https://github.com/sl4m3/ledgermind/releases/latest/download/install.sh | sh
bash
curl -fsSL https://github.com/sl4m3/ledgermind/releases/latest/download/install.sh \
  | sh -s -- install --non-interactive \
      --config /secure/ledgermind-install.json --json

Technical Analysis

The installation instructions retrieve a shell script from a mutable latest GitHub release URL and pipe it directly into sh. The downloaded script is not included in the audited package, so its behavior cannot be reviewed as part of this Skill. Its content can also change after the Skill has passed an audit.

Although the documentation states that installation should proceed only when a signed 4.0 release is available, the supplied commands do not perform cryptographic signature or checksum verification. They also do not pin an immutable release version, commit, or artifact digest. Merely using HTTPS and hosting the artifact on GitHub does not establish the integrity of the publisher account or future release artifacts.

The non-interactive command additionally supplies the remote installer with a private configuration file path. Even if that file follows the documentation and contains only credential references rather than plaintext tokens, an untrusted installer executes with the user's authority and can read other files accessible to that user.

Executing installation code is necessary for the declared installation functionality, but immediately executing an unverified, mutable remote payload is not the minimum safe mechanism required to achieve that functionality.

Attack Path

  1. An attacker compromises the repository owner account, release publication process, or release artifact.
  2. The attacker replaces ...[truncated 1479 chars]
Remediation
View remediation

Remediation Suggestions

  1. Replace the mutable latest URL with an explicitly pinned, immutable release version.
  2. Download the installer to a local file instead of piping it directly into a shell:
    bash
    curl -fL --output ledgermind-install.sh \
      https://github.com/sl4m3/ledgermind/releases/download/VERSION/install.sh
    
  3. Publish a SHA-256 digest and a cryptographic signature for the installer through a trusted release process.
  4. Verify both the digest and signature before execution, using a signing key whose fingerprint is documented through an independent trusted channel.
  5. Abort installation if any verification fails. Do not silently fall back to another release or unsigned artifact.
  6. Allow the user or agent to inspect the downloaded script before explicitly approving its execution.
  7. Run installation with the least-privileged account capable of completing the task. Do not recommend sudo unless a specific operation demonstrably requires it.
  8. Document the exact files, integrations, services, and permissions the installer will create or modify.
  9. Avoid exposing configuration to the installer until its integrity has been verified. Keep configuration permissions restricted and continue prohibiting plaintext credentials.
  10. Prefer distributing a reproducible, versioned package through a trusted package mechanism where artifact provenance and integrity can be validated.
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Chaining Abuse

High
Category
Tool Misuse
Confidence
99% confidence
Finding

The | sh construct is the dangerous core of the installation instruction because it turns network-delivered content into immediate shell execution. In a discovery/install skill, this context makes the risk more serious: users may trust the guidance and run it verbatim, enabling one-step remote code execution if anything in the delivery chain is tampered with.

Content

Scanner excerpt · SKILL.md (reported line 68)May include surrounding context.

If a signed 4.0 release is available, the interactive installer is:

bash
curl -fsSL https://github.com/sl4m3/ledgermind/releases/latest/download/install.sh | sh

For an agent-assisted installation, create a private configuration from the

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
80% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · SKILL.md (reported line 75)May include surrounding context.

user's choices and run:

bash
chmod 600 /secure/ledgermind-install.json
curl -fsSL https://github.com/sl4m3/ledgermind/releases/latest/download/install.sh \
  | sh -s -- install --non-interactive \
      --config /secure/ledgermind-install.json --json

External Script Fetching

Low
Category
Supply Chain
Confidence
98% confidence
Finding

The skill instructs users to fetch a remote shell script and immediately execute it via curl ... | sh, which removes the opportunity to inspect or verify the script contents before execution. If the release endpoint, repository, maintainer account, or network path is compromised, this leads to arbitrary code execution on the host.

Content

Scanner excerpt · SKILL.md (reported line 68)May include surrounding context.

If a signed 4.0 release is available, the interactive installer is:

bash
curl -fsSL https://github.com/sl4m3/ledgermind/releases/latest/download/install.sh | sh

For an agent-assisted installation, create a private configuration from the

External Script Fetching

Low
Category
Supply Chain
Confidence
98% confidence
Finding

This non-interactive installer variant also fetches a remote script and executes it directly, while additionally passing a path to a private configuration file into the installer. A compromised script could execute arbitrary commands and potentially read or exfiltrate installer configuration or referenced secrets.

Content

Scanner excerpt · SKILL.md (reported line 76)May include surrounding context.

bash
chmod 600 /secure/ledgermind-install.json
curl -fsSL https://github.com/sl4m3/ledgermind/releases/latest/download/install.sh \
  | sh -s -- install --non-interactive \
      --config /secure/ledgermind-install.json --json