logmind
Fail
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONPERSISTENCECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill's setup instructions suggest installing the tool via
curl -fsSL https://logmind.dev/install.sh | bash. Executing remote scripts directly in a shell is a dangerous practice that can lead to arbitrary code execution from a third-party source without manual inspection. - [REMOTE_CODE_EXECUTION]: The skill documentation mentions using
npx clud-bug@latest init, which fetches and executes code from the official npm registry at runtime. - [PERSISTENCE]: The
logmind initcommand installs local Git hooks (pre-commit,post-merge) and GitHub Action workflows (regen-timeline.yml,check-decisions.yml). This establishes persistence by ensuring the tool's logic is automatically triggered during standard development workflows. - [COMMAND_EXECUTION]: The skill makes extensive use of a local CLI tool to modify the project's working tree and Git history. Specifically,
logmind logis designed to stage the entire working tree and create commits automatically. - [COMMAND_EXECUTION]: The
logmind skill pushcommand is described as cloning a catalog and opening a real pull request without a confirmation prompt, which could lead to unintended external modifications. - [INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it is instructed to ingest data from project-level documentation and decision logs (e.g.,
docs/timeline.md,docs/decisions.md). This creates an attack surface where malicious instructions embedded in those files could influence the agent's behavior. - Ingestion points: Reads from various markdown files in the
docs/directory to gain project context. - Boundary markers: None mentioned; the agent is directed to treat the content of these files as authoritative instructions for its task.
- Capability inventory: The agent has the ability to execute shell commands, modify files, and interact with Git via the
logmindtool. - Sanitization: No sanitization or validation of the markdown content is described before the agent processes it.
Recommendations
- HIGH: Downloads and executes remote code from: https://logmind.dev/install.sh - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata