The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the installation of the @bitkyc08/opencodex package from the public NPM registry, which is an external dependency from an unverified source.
[PERSISTENCE]: The skill provides instructions for configuring background services and launcher shims that ensure the proxy service automatically restarts and persists across user login sessions.
[DATA_EXFILTRATION]: The skill accesses and manages sensitive configuration files located at ~/.opencodex/config.json and ~/.opencodex/service-api-token, which store API keys and management authentication tokens.
[COMMAND_EXECUTION]: The skill utilizes system commands to modify global environment settings using mise and to perform global package installations via npm install -g, affecting the host system configuration.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes operational logs (service.log and usage.jsonl) that may contain data from external requests, creating a potential surface for indirect prompt injection if malicious content is processed from these logs.
Ingestion points: Reads ~/.opencodex/service.log and ~/.opencodex/usage.jsonl (SKILL.md).
Boundary markers: No explicit delimitations or instructions to ignore embedded content were identified.
Capability inventory: The skill executes npm install, ocx CLI utilities, and complex shell pipelines (SKILL.md).
Sanitization: There is no evidence of filtering or sanitizing log entries before the agent interprets their content.

opencodex-setup