The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill uses the !command`` syntax in SKILL.md to execute the scripts/probe-els.sh script when the skill is loaded. This script gathers environment metadata by reading local configuration files like package.json and tsconfig.json to identify the version and patch status of the Effect Language Service. This is a benign use of dynamic context for environmental discovery.
[COMMAND_EXECUTION]: The instructions guide the agent to perform various command-line operations via Bash, including running effect-language-service diagnostics, quickfixes, and codegen. It also includes a patch command that modifies the local node_modules/typescript installation to integrate Effect diagnostics. These operations are consistent with the skill's primary purpose and represent standard developer workflows.
[INDIRECT_PROMPT_INJECTION]: The skill processes output from CLI tools that may reflect content from source code, such as diagnostics on code comments. This represents a potential surface for indirect prompt injection. Ingestion points: CLI output from diagnostics, quickfixes, overview, and layerinfo commands. Boundary markers: Absent in the provided workflows. Capability inventory: Access to Bash, Read, Glob, and Grep. Sanitization: The els_clean helper function strips ANSI formatting but does not validate the content of the tool's output.

effect-language-service-cc