The Agent Skills Directory

[SAFE]: The skill provides purely instructional and methodological content for telemetry design without including scripts, executables, or commands.
[EXTERNAL_DOWNLOADS]: The skill references an evaluation JSON file and a canonical skill path on the author's GitHub repository (github.com/jacob-balslev/skill-graph). These are informational references to vendor-owned documentation and do not involve runtime execution of remote code.
[COMMAND_EXECUTION]: The skill restricts its environment using the allowed-tools frontmatter field to Read and Grep, which are passive information-gathering tools. No shell command execution or privileged operations were found.
[DATA_EXFILTRATION]: There are no patterns suggesting the collection or transmission of sensitive data. The methodology explicitly includes a step to "Define privacy redaction" and ensures "Sensitive data is redacted before emission."
[PROMPT_INJECTION]: The instructions are consistent with the stated purpose of observability modeling and do not contain directives to bypass safety filters or ignore system instructions.

observability-modeling