The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to fetch documentation from external URLs (e.g., docs.deep.space/llms.txt, docs.deep.space/llms-full.txt) and explicitly designates this external content as the "authority for everything else." This creates a surface where external content can override agent behavior.
Ingestion points: Documentation pages fetched via HTTP/curl from https://docs.deep.space/ (referenced in SKILL.md).
Boundary markers: Absent; the skill explicitly directs the agent to adopt the opinionated approach of the external documentation as the default.
Capability inventory: Access to npx deepspace CLI which handles authentication, deployment, secret management, and ownership transfers.
Sanitization: No sanitization or validation of the external documentation content is mentioned before it is processed by the agent.
[COMMAND_EXECUTION]: The skill relies on the execution of multiple CLI commands via npx deepspace for core operations including authentication, project scaffolding, secret storage, and deployment.
npx deepspace auth login: Handles OAuth authentication.
npx deepspace secrets: Manages application secrets.
npx deepspace deploy and npx deepspace app transfer: Manages application lifecycle and ownership.
[EXTERNAL_DOWNLOADS]: The skill facilitates the download of external code and configurations.
npm create deepspace@latest: Downloads and executes a remote project scaffold.
Documentation retrieval via curl -fsSL or bounded HTTP requests targeting docs.deep.space.
[CREDENTIALS_UNSAFE]: The skill manages sensitive operations like authentication and secret storage, although it provides specific instructions to mitigate risks.
Mentions deepspace secrets for storage instead of .dev.vars or environment variables.
Instructs the agent never to handle passwords directly and to use operator-supplied environment credentials in CI environments.

deepspace