The Agent Skills Directory

[Indirect Prompt Injection Surface]: The skill is designed to ingest and process data from external sources, specifically AWS API outputs and CloudWatch log events, which could theoretically contain instructions intended to influence the agent's behavior.
Ingestion points: Data enters the agent's context through AWS API calls such as logs:FilterLogEvents, logs:GetLogEvents, and various Describe or Get calls to the Bedrock AgentCore and X-Ray services, as defined in SKILL.md and references/checks-catalog.md.
Boundary markers: The instructions do not explicitly require the use of boundary markers (e.g., XML tags or delimiters) when processing these external data points.
Capability inventory: The skill utilizes read-only AWS service APIs to verify configuration states and performs file-write operations to generate a report artifact.
Sanitization: There is no explicit instruction for the agent to sanitize or filter the content of log events before including them in the final report. This is a common characteristic of diagnostic tools and is considered low risk in the context of official AWS service metadata.
[IAM Permission Recommendations]: The skill provides example IAM policy documents in references/iam-tiers.md to help users grant the necessary read-only permissions.
Context: The examples use Resource: "*" for several read-only actions (e.g., logs:DescribeLogGroups). While broad, the documentation includes a 'Scoping note' advising users to restrict these permissions to specific ARNs for tighter security, representing a standard security trade-off for accessibility in diagnostic samples.
[Processing of Service Logs]: The skill validates observability by checking for the arrival of telemetry in log streams (logs:FilterLogEvents).
Context: While this exposes the content of service logs to the agent, this is a functional requirement for the skill to confirm that spans and metrics are flowing correctly. The skill is read-only and does not exfiltrate this data beyond the local report artifact.

agentcore-observability-setup