The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill is designed to manage sensitive files and credentials. It interacts with paths such as .env, ~/.ssh/id_ed25519, and cloud provider configuration stores to perform lifecycle operations like rotation and provisioning.
[COMMAND_EXECUTION]: The skill uses various command-line tools including gh, aws, gcloud, vercel, fly, netlify, and railway to interact with remote secret stores and deployment environments.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection as it ingests and processes environment configuration data.
Ingestion points: Reads and audits .env and .env.example files (referenced in SKILL.md and scripts/check-env-drift.sh).
Boundary markers: Utilizes a required "Assumptions block" to scope the agent's actions and environment tier.
Capability inventory: Has permissions for file writing, editing, and executing shell commands via cloud CLIs.
Sanitization: Employs fingerprinting for secret comparison and mandates the use of placeholders to prevent accidental exposure of raw credentials.

manage-secrets-env