The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses npx -y @weekly-git-report/agent-cli@latest to execute core functionality. This pattern downloads and runs code from the NPM registry at runtime, allowing the execution of arbitrary code that is not part of the skill's reviewed source.
[EXTERNAL_DOWNLOADS]: The skill triggers the download of the @weekly-git-report/agent-cli package. Since it targets the @latest version tag, it is vulnerable to supply chain attacks where a malicious update to the package would be automatically executed.
[COMMAND_EXECUTION]: The skill requires the agent to execute several shell commands to sync projects, collect Git logs, and save summaries. This provides a broad attack surface if the underlying CLI tool or the inputs provided to it are compromised.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes raw Git commit history, which is untrusted external data.
Ingestion points: The agent reads Git commit data via the raw read command in SKILL.md.
Boundary markers: While the skill instructs the agent to only use raw records and not invent content, there are no technical delimiters or specific instructions to ignore embedded commands within the commit messages.
Capability inventory: The agent has the capability to execute shell commands (via npx) and write files to the system.
Sanitization: There is no evidence of sanitization or filtering applied to the commit messages before they are processed by the agent for summarization.

weekly-git-report