The Agent Skills Directory

[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill uses uvx to download and execute code directly from git+https://github.com/ambient-code/agentready. This execution occurs without prior verification of the repository's contents or author, granting the external code the ability to perform operations within the agent's environment. This pattern is repeated during both the initial assessment phase and the post-remediation verification step.\n- [EXTERNAL_DOWNLOADS]: The skill fetches the agentready package from a third-party GitHub organization (ambient-code) that is not identified as a trusted vendor. This creates a runtime dependency on an unverified external source.\n- [COMMAND_EXECUTION]: The skill performs shell commands to interact with git for worktree validation and the uvx tool for assessment. It interpolates repository paths and configuration file paths provided by the user into these commands, which presents a risk of command injection if paths are specially crafted.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from multiple sources within the target repository and from the external tool's output.\n
Ingestion points: Reads assessment-latest.json (produced by the external tool), as well as package.json, Makefiles, pyproject.toml, and CI workflow files from the target repository.\n
Boundary markers: The instructions do not define boundary markers or 'ignore' instructions when processing this external content.\n
Capability inventory: The skill has the capability to write new files (AGENTS.md, CLAUDE.md) to the repository and execute shell commands/remote code via uvx.\n
Sanitization: There is no evidence of sanitization or filtering of the content extracted from the external files before it is used to influence the agent's actions or written to new files.

agent-readiness