ACB-GEN-003: Artifacts contain no embedded secrets
| Field | Value |
|---|---|
| Status | draft |
| Severity | high |
| Check type | static |
| Applies to | skill, instruction-file, mcp-server-declaration |
| Profiles | strict-general |
| Exceptions | not-permitted |
Description
No artifact contains credentials, tokens, private keys or connection strings with embedded passwords. MCP server declarations reference secrets through environment variable indirection or a secret store, never as literal values.
Rationale
Agent artifacts are designed to be shared, forked and published to registries, and their contents are loaded into model context where they can be echoed, logged or exfiltrated. A secret in an artifact is a secret disclosed to every consumer and every agent session that reads it.
Audit procedures
secret-scan (pattern)
Run a secret scanner with high-confidence provider rules and generic high-entropy detection over all files. Any finding fails.
Illustrative failing indicators (non-exhaustive):
-----BEGIN OPENSSH PRIVATE KEY-----postgres://admin:hunter2@db.example.com/prod
mcp-env-indirection (structure)
In MCP server declarations, values of `env`, `headers` and argument fields whose key or flag name suggests a credential (token, key, secret, password, auth) are environment-variable references or placeholders, not literals.
Illustrative failing indicators (non-exhaustive):
"env": {"API_TOKEN": "sk-live-..."}"Authorization": "Bearer eyJ..."
Remediation
Remove the secret, rotate it, and purge it from version history. Replace literals with environment variable references and document the variable name the consumer must provide.