The Agent Skills Directory

[COMMAND_EXECUTION]: The skill interacts with the local file system to execute custom scripts located in the ~/Projects/manager/bin directory, such as namecheap-set-ns and cloudflare-ai-bots, as part of its core functionality.
[CREDENTIALS_UNSAFE]: The instructions direct the agent to source sensitive environment variables, including CLOUDFLARE_API_TOKEN and CF_API_TOKEN, from standard shell profiles (~/.profile) or project-specific configuration files. This follows established patterns for local developer tooling.
[DATA_EXPOSURE]: The skill reads from local Markdown files (DOMAINS.md, DNS.md, redirect-worker-mapping.md) to determine the state and intended configuration for domain management operations.
[PROMPT_INJECTION]: There is a surface for indirect prompt injection because the agent treats the content of files within ~/Projects/manager as the authoritative source of truth. If these local files were compromised, they could potentially influence the agent's actions during execution.
Ingestion points: ~/Projects/manager/DOMAINS.md, ~/Projects/manager/DNS.md, ~/Projects/manager/redirect-worker-mapping.md.
Boundary markers: None identified; the agent is instructed to use these files as primary instructions.
Capability inventory: Execution of cli4 commands, execution of local binaries in bin/, and Git operations (commit/push).
Sanitization: None identified; the skill relies on the integrity of the local project files.

domain-dns-ops