Caveat verdict
atomicmail
Accesses credentials AND encodes data
The skill is an email client using JMAP against documented atomicmail.ai endpoints with credentials stored in local credential files; all behavior matches the stated purpose of reading and writing email through a legitimate mail service.
Automated static analysis — not a human review. Caveat flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
What it does
These are capability combinations: each listed behavior occurs in the skill, but Caveat detects co-occurrence — it does not verify that one flows into another. Read the code to confirm a live chain.
Accesses credentials AND encodes data — may obfuscate stolen credentials
LLM02 · ASI03 · ASI04
Permission integrity
credential_access
Findings (6)
Uses exec() — may execute shell commands
lib/esm/lib/agent/auth/agent-auth-http.js · prose · downgraded · exec(
Base64 decode (atob) — may hide malicious payloads
lib/esm/lib/agent/auth/agent-jwt.js · prose · downgraded · atob(
Accesses system credential store
lib/esm/lib/agent/session/agent-credentials-store.d.ts · prose · downgraded · CredentialStore
Long base64 string (100+ chars) — likely obfuscated payload
lib/shared/fixtures/pow_vectors.json · prose · downgraded · 0e365cf9d7a9a9d515b5362da1ed60b3c287d6b5de7f5f0e55b15bd8a34dd3605695e20f01fb60fb
References agent configuration files
lib/esm/lib/agent/session/agent-resolve-config.d.ts · prose · downgraded · AgentConfig
Accesses sensitive environment variables
lib/esm/lib/agent/session/agent-resolve-config.js · prose · downgraded · process.env.ATOMIC_MAIL_CREDENTIAL
Permissions & capabilities
No declared permissions — minimal attack surface.
credential_accessdata_encoding Thanks — recorded.