ℹ
Purpose & Capability
The name/description align with the runtime instructions: the SKILL.md shows CLI commands to get an agent email, read inbox, and store/retrieve encrypted passwords. Requiring a 'sunday' binary is consistent. However, the skill writes sensitive state to ~/.sunday/config.json (stores credentials/encryption keys) yet the registry metadata lists no required config paths or primary credential — an inconsistency that should have been declared.
⚠
Instruction Scope
SKILL.md instructs interactive login (opens browser, enter 6‑digit PIN) and then states that credentials and encryption keys are stored in ~/.sunday/config.json so subsequent commands run autonomously. The instructions give the agent explicit commands to list inboxes, fetch messages, and decrypt passwords. This is functionally expected for the stated purpose, but it grants autonomous access to highly sensitive data (email, OTPs, passwords) and the file path for those secrets is not declared in the skill manifest.
⚠
Install Mechanism
Install is via a Homebrew formula from a third‑party tap (ravi-technologies/tap/sunday). Homebrew itself is a common channel, but the tap is not a widely-known/official upstream and no homepage/source repo is provided in the metadata — you should verify the formula and upstream source before installing. The install type is not an arbitrary URL download, which reduces risk, but trust in the tap is unclear.
⚠
Credentials
The skill declares no required environment variables, but it stores encryption keys/credentials client-side in ~/.sunday/config.json and promises no further PIN prompts after initial login. That means the agent (and any process with access to that file) can access the vault persistently. The manifest did not declare this config path or any primary credential, so the degree of persistent secret access is understated in the metadata.
ℹ
Persistence & Privilege
The skill is not force-installed (always:false) and model invocation is allowed (default). Autonomous invocation combined with stored local keys yields a higher blast radius because the agent can autonomously read inboxes and decrypt passwords without human re‑authentication. This combination is expected for this class of skill but is sensitive — verify you trust the binary and its install location.