The Agent Skills Directory

[DATA_EXFILTRATION]: Operational protocols require the agent to request sensitive system context, including cloud provider metadata and target environment configurations. (SKILL.md)
[DATA_EXFILTRATION]: Multiple reference documents contain extreme technical noise, specifically 150 near-identical sections per file, which functions as an audit-evasion technique to complicate manual and automated auditing. (references/architecture-patterns.md, references/code-organization.md, references/deployment-pipelines.md, references/error-handling.md, references/performance-optimization.md, references/security-best-practices.md, references/state-management.md, references/testing-strategies.md)
[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection through the ingestion of external code and configuration data. 1. Ingestion points: target source code and environment configurations (SKILL.md). 2. Boundary markers: Absent from the instructions. 3. Capability inventory: auditing logic and finding reporting (SKILL.md). 4. Sanitization: generic instruction to sanitize inputs is present but lacks specific mechanisms for ingested file contents (SKILL.md).

iam-governance