The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides instructions for the agent to install the barrage tool by downloading a shell script from the vendor's GitHub repository and piping it directly to bash (curl -fsSL https://raw.githubusercontent.com/codetesla51/barrage/main/install.sh | bash).
[EXTERNAL_DOWNLOADS]: The skill fetches installation scripts and Go packages from GitHub repositories belonging to the vendor (codetesla51/barrage).
[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands, including go build, go run, and the barrage CLI itself, to perform diagnostics and seed databases.
[INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to ingest and process local project data (README, schema, migrations, .env files) to extract configuration values for the load tests.
Ingestion points: Defined in the 'Inspect the CURRENT project first' section of SKILL.md, where the agent is told to read README, routes/handlers, DB migrations/schema, and configuration files.
Boundary markers: No specific delimiters or safety warnings are instructed for the data being read.
Capability inventory: The agent has the ability to execute shell commands (barrage run, go build, etc.) and perform network operations via the tool.
Sanitization: The skill includes behavioral instructions to "Never assume URLs, ports, DB conns... derive them from the project or ask," but lacks programmatic sanitization of the input file content.

barrage