Credentials in tool parameters
A tool call contains a value that resembles an API key, token, or password.
Why it is risky
Section titled “Why it is risky”Passing a secret through general-purpose tools can expose it to an unintended service, process, or recipient. The impact depends on the secret permissions and where the tool sends or stores it.
Example
Section titled “Example”An agent pastes a service token into the body of a support-ticket tool instead of referring to the approved secret store.
This is a fictional example, not an event from a customer.
What Omnodex Detects
Section titled “What Omnodex Detects”The rule scans captured parameters for supported credential formats and labeled values. It filters recognizable placeholders and selected documentation examples. It does not verify that a credential is valid.
| Rule ID | Event category | Tier | Base severity |
|---|---|---|---|
RULE_CREDENTIAL_IN_PARAMS | credential_exposure | Community | MEDIUM |
Base severity is the rule default. A finding involving content saved for later execution may have a lower severity. Your loaded rule version and event details determine the finding you see.
What to check next
Section titled “What to check next”Identify the recipient and whether the value was real. Remove the unnecessary disclosure and rotate a credential if it reached an unauthorized recipient. Use the tool or service secret-handling mechanism for future calls.
False positives
Section titled “False positives”A realistic synthetic token or an authorized authentication parameter may trigger the rule. Test credentials can also be sensitive even when they cannot access production.
False negatives
Section titled “False negatives”Unknown formats, encoded or split values, short tokens, and real credentials that resemble filtered placeholders may be missed. Values absent from the captured parameters cannot be inspected.
Technical details
Section titled “Technical details”This is a single-call pattern finding with a base severity of MEDIUM. Matching type labels help identify what to review; they do not establish that an account was accessed.
A finding describes a pattern in captured activity. Review the original call, its result, and the task context before concluding that harm occurred. Absence of a finding does not establish that an operation is safe.