The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches release information and changelog data from the official Anthropic GitHub repository (https://api.github.com/repos/anthropics/claude-code/releases and https://raw.githubusercontent.com/anthropics/claude-code/main/CHANGELOG.md). These are trusted sources and the operations are documented neutrally as intended functionality.
[COMMAND_EXECUTION]: The skill utilizes bash to run gh api, curl, and date commands. These commands are used to interact with GitHub APIs and perform date calculations. The implementation does not interpolate unsanitized user input into these shell commands.
[PROMPT_INJECTION]: As an indirect prompt injection surface, the skill ingests external data from release bodies.
Ingestion points: Data enters the agent context through the GitHub API and CHANGELOG.md file content.
Boundary markers: None explicitly defined in the output formatting.
Capability inventory: The skill uses Bash and WebFetch for data retrieval and performs file writes to the local directory to save the report.
Sanitization: None detected for the external content before formatting.
Note: While this is a vulnerability surface, the data source is an official repository of a trusted organization, making the practical risk negligible in this context.

claude-changelog