The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The SKILL.md file contains instructions for the agent to execute 'python3 scripts/install-external-skills.py' to add capabilities such as 'deslop' or 'improve'. This script is not present in the provided skill files, resulting in the execution of unverifiable code retrieved from an external registry.
[EXTERNAL_DOWNLOADS]: The installation process for external skills relies on registry and source files (external-skills.yaml and external-sources.yaml) located outside the skill's local directory, which implies the retrieval of components from remote or external sources.
[COMMAND_EXECUTION]: The skill includes 'scripts/find_latest_release_tag.sh', a shell script that interacts with the git CLI to fetch and list tags. This provides the agent with a mechanism to execute shell commands and communicate with remote repositories.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process potentially untrusted external data such as CI failure logs, test outputs, and git diffs. Ingestion points: CI failure reports and git diffs as described in SKILL.md and release-review-checklist.md. Boundary markers: No delimiters or explicit instructions are provided to the agent to treat this ingested data as untrusted. Capability inventory: Access to shell script execution, git operations, and the ability to trigger external skill installations. Sanitization: No sanitization or validation logic is defined for the ingested data before it is processed by the agent.

flake