The Agent Skills Directory

[METADATA_POISONING]: The skill provides systematic misinformation regarding the current versions and technical architectures of several industry-standard tools. It describes a non-existent Go-native rewrite for 'TypeScript 7', references future-dated versions of Rust (1.97), Vite (8), and ESLint (10), and claims 'Prisma 7' is a breaking release. While these claims are framed as technical guidance, this deceptive data could lead an agent to attempt to install non-existent or malicious versions of these tools if version-squatting occurs on public registries.
[INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to ingest and process data from the user's repository, including documentation files and arbitrary code files discovered through codebase exploration. This ingestion of untrusted content combined with the agent's ability to execute commands and spawn sub-agents represents an indirect prompt injection attack surface.
Ingestion points: Files ARCHITECTURE.md, CONTEXT.md, and codebase modules scanned by the Explore sub-agent (SKILL.md).
Boundary markers: Absent; the skill does not specify the use of delimiters or warnings to ignore embedded instructions in ingested files.
Capability inventory: The skill is instructed to execute shell commands for dependency management and project tooling, and it utilizes the Agent tool to spawn sub-agents for architectural exploration and interface design.
Sanitization: No evidence of escaping or validation of ingested content before processing.
[COMMAND_EXECUTION]: The skill's tooling convention files (PYTHON.md, RUST.md, TYPESCRIPT.md, PRISMA.md) provide instructions for the agent to execute various shell commands for dependency management and code analysis, including uv run, cargo build, pnpm install, and npx prisma.

architecture