The Agent Skills Directory

Unverifiable Dependencies (MEDIUM): The skill installation instructions recommend npx skills add axiomhq/skills. Because axiomhq is not included in the list of trusted organizations, this represents a download and execution of software from an unverified source.
Indirect Prompt Injection (LOW): The skill analyzes data from the axiom-audit and axiom-history datasets, which are external sources that could contain malicious instructions embedded in log data.
Ingestion points: Audit logs and query history data processed by scripts/baseline-stats and scripts/analyze-query-coverage.
Boundary markers: Absent; no delimiters or "ignore instructions" warnings are present in the scripts or the SKILL.md instructions for processing log content.
Capability inventory: Access to axiom-api (network), and capabilities to create/update/delete dashboards and monitors.
Sanitization: The skill uses jq for structure parsing but lacks specific sanitization for string content before it is interpolated into analysis results or displayed to the agent.

controlling-costs