From fdebb067200ac4df3baeb82a9aaffdea0539d27e Mon Sep 17 00:00:00 2001 From: Jeroen Willemsen Date: Mon, 26 Oct 2026 08:45:42 +0100 Subject: [PATCH] adding remote secret proxy pattern --- src/main/resources/explanations/challenge69_reason.adoc | 1 + src/main/resources/explanations/challenge72_reason.adoc | 1 + src/main/resources/explanations/challenge73_reason.adoc | 1 + 3 files changed, 3 insertions(+) diff --git a/src/main/resources/explanations/challenge69_reason.adoc b/src/main/resources/explanations/challenge69_reason.adoc index e62227c374..cde013cba1 100644 --- a/src/main/resources/explanations/challenge69_reason.adoc +++ b/src/main/resources/explanations/challenge69_reason.adoc @@ -16,6 +16,7 @@ What to do instead: - Give every consumer their own short-lived, least-privilege credential instead of one shared team token. - Scan skill files with a secret scanner in pre-commit and CI, just like you scan source code. - Treat a secret that ever appeared in a skill as compromised and rotate it. +- In case of (autonomous and/or remote) agents: always use something like a secrets proxy pattern. ---- [NOTE] diff --git a/src/main/resources/explanations/challenge72_reason.adoc b/src/main/resources/explanations/challenge72_reason.adoc index 54bc8a7fcb..4959ff7ee8 100644 --- a/src/main/resources/explanations/challenge72_reason.adoc +++ b/src/main/resources/explanations/challenge72_reason.adoc @@ -24,6 +24,7 @@ What to do instead: - Review agent transcripts before saving or sharing them, just as you would review a pull request. - Treat any transcript that read from a secrets source as potentially compromised, and rotate the exposed credentials. - Use short-lived, scoped tokens for staging deployments so that exposure has limited blast radius. +- In case of (autonomous and/or remote) agents: always use something like a secrets proxy pattern. ---- [NOTE] diff --git a/src/main/resources/explanations/challenge73_reason.adoc b/src/main/resources/explanations/challenge73_reason.adoc index ca47d928b6..8b9b9121f8 100644 --- a/src/main/resources/explanations/challenge73_reason.adoc +++ b/src/main/resources/explanations/challenge73_reason.adoc @@ -21,6 +21,7 @@ What to do instead: - Keep credentials out of the instruction files and let the agent read them from the environment, for example `$STAGING_DEPLOY_TOKEN`. +- In case of (autonomous and/or remote) agents: always use something like a secrets proxy pattern. - Document *which* variable is required and *where* to obtain it, never the value itself. - Give every consumer a short-lived, least-privilege credential instead of one