The Agent Skills Directory

[PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection by instructing the agent to read and respect local project context files.
Ingestion points: Reads project-specific constraints from .agents/qa-project-context.md to guide environment design.
Boundary markers: No specific delimiters or instructions to ignore embedded commands within the context file are provided.
Capability inventory: The skill utilizes shell-based capabilities including docker compose, npm, npx, psql, and curl to manage infrastructure and data.
Sanitization: No explicit sanitization or validation of the ingested context data is mentioned before it influences command execution.
[DATA_EXPOSURE]: The skill explicitly warns against using production database copies in test environments to prevent PII exposure and recommends factory-based or anonymized data seeding.
[COMMAND_EXECUTION]: Provides standard devops commands for container orchestration and database management. These are well-documented and necessary for the skill's intended purpose.
[SAFE]: All external tools, images (PostgreSQL, Redis, MinIO, MailHog), and libraries (MSW, Playwright, Vitest) referenced in the examples are standard, well-known resources in the software development ecosystem.

test-environments