The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches the configuration schema from the author's GitHub repository (TitusKirch/skills) using curl. This download is used to provide up-to-date validation and documentation for configuration keys.
[COMMAND_EXECUTION]: The agent is instructed to run a local shell script (templates/resolve-config.sh) to merge configuration profiles. It also executes standard commands such as git and gh (GitHub CLI) to inspect the repository state, remotes, and authentication status. These operations are within the scope of the skill's purpose.
[DATA_EXFILTRATION]: The skill reads repository metadata including git remotes, issue templates, and labels. This data is used locally to propose configuration defaults and identify drift (e.g., templates referencing missing labels). The skill only writes to the local .tituskirch-skills.json file and does not transmit data to untrusted external domains.
[PROMPT_INJECTION]: The skill ingest untrusted data from repository issue templates (.github/ISSUE_TEMPLATE/) to check for label consistency. This represents a potential surface for indirect prompt injection. However, the skill implements strong guardrails, including mandatory user confirmation for all plans and writes, and validation against a strict JSON schema, which mitigates the risk of malicious instructions influencing the agent's behavior.

tituskirch-skills-config