The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute several local development and diagnostic commands, such as go build, go vet, diff, awk, sed, and jq. It also invokes the vendor-provided cli-printing-press binary to manage build locks, perform dogfooding, and run verification suites.
[EXTERNAL_DOWNLOADS]: The skill fetches and installs the cli-printing-press binary from the vendor's official GitHub repository (mvanhorn/cli-printing-press). It also downloads the gosec security scanner from its well-known repository (securego/gosec) if it is not already present on the local system.
[REMOTE_CODE_EXECUTION]: The skill executes code from remote sources using go install for the vendor's tool and go run for the well-known gosec security scanner. These operations are standard setup and diagnostic procedures for development environments.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the CLI project directory being polished, including Go source code, manifests, and research files, which presents a surface for indirect prompt injection.
Ingestion points: Reads project files like SKILL.md, research.json, .printing-press.json, tools-manifest.json, and Go source code to inform its fixing logic.
Boundary markers: Employs a specialized sub-skill (printing-press-output-review) to perform agentic review of generated outputs.
Capability inventory: Has permissions to write and edit project files and execute local shell commands via Bash.
Sanitization: Implements diagnostic tools such as gosec for security scanning and pii-audit to detect and remove sensitive customer data from the processed content.

printing-press-polish