The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on shell commands to inspect the repository and interact with APIs. Evidence includes git remote -v, gh issue list, gh pr view, and glab issue list found in SKILL.md and the issue-tracker templates. These commands are executed to discover repository configuration and manage ticket workflows.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8) by reading and processing content from external, user-controllable sources such as GitHub Issues and GitLab Merge Requests. 1. Ingestion points: The agent is instructed to use gh issue view <number> --comments and glab issue view <number> --comments in issue-tracker-github.md and issue-tracker-gitlab.md respectively. 2. Boundary markers: The provided templates do not specify explicit delimiters or instructions for the agent to ignore or escape potentially malicious content embedded in issue bodies or comments. 3. Capability inventory: The agent has the capability to write to the file system (e.g., creating AGENTS.md, docs/agents/*.md) and execute further CLI commands based on the processed information. 4. Sanitization: The instructions lack specific validation or sanitization steps for content fetched from remote issue trackers.
[CREDENTIALS_UNSAFE]: The skill explores the .git/config file during its exploration phase in SKILL.md. While intended for metadata discovery (such as identifying the repository URL), this file can occasionally contain plaintext authentication tokens or sensitive credentials depending on the user's local configuration, which could lead to accidental data exposure in the agent's context.

setup-matt-pocock-skills