The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the installation of the @aicontextlab/cli package via npm and references resources from the 0xranx GitHub repository and GitHub Pages domains.
[COMMAND_EXECUTION]: The skill makes extensive use of shell commands, including global package installation (npm install -g), dynamic execution via npx, and interaction with the oc CLI tool for project initialization and context management.
[INDIRECT_PROMPT_INJECTION]: The skill processes external documents and context files to provide memory for the agent. This creates an attack surface where malicious instructions embedded in processed documents could influence agent behavior.
Ingestion points: Document search (oc_search, oc search), folder listing (oc_list_folders), and manifest generation (oc_manifest, oc context manifest) described in SKILL.md.
Boundary markers: No explicit delimiters or safety warnings for ingested content are defined in the instructions to prevent the agent from executing instructions found within the context files.
Capability inventory: The skill is permitted to use Bash, Write, Read, Grep, and Glob tools across the filesystem.
Sanitization: There is no evidence of sanitization, escaping, or filtering of the retrieved document content before it is interpolated into the agent's context.
[CREDENTIALS_UNSAFE]: The instructions include commands to set and store sensitive embedding API keys locally using the oc config set EMBEDDING_API_KEY command. While the examples use placeholders, the pattern encourages the storage of secrets in the tool's local configuration.

opencontext