The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted local data (Indirect Prompt Injection surface).
Ingestion points: The skill reads package.json, pyproject.toml, go.mod, Cargo.toml, and .project-standards.yaml to detect project settings.
Boundary markers: There are no explicit instructions or delimiters used to prevent the agent from following instructions that might be embedded in these files (e.g., in a package description or script name).
Capability inventory: The skill has access to Bash, Write, Edit, and WebFetch tools, which could be exploited if malicious instructions in project files were followed.
Sanitization: The skill uses standard tools like jq and grep to extract data but does not explicitly sanitize the content before the agent processes it.
[EXTERNAL_DOWNLOADS]: The skill downloads several software development kits (SDKs) and providers.
Trusted Sources: All downloads originate from well-known and trusted registries: NPM (@openfeature/*), PyPI (openfeature-sdk), and GitHub (github.com/open-feature/*, github.com/thomaspoignant/*). Per security guidelines, these are documented neutrally and do not escalate the verdict.
[COMMAND_EXECUTION]: The skill uses Bash to perform local setup tasks such as installing packages, linting configuration files, and checking service health on localhost. These operations are consistent with the skill's primary purpose.

configure-feature-flags