The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill contains explicit instructions to bypass platform security boundaries. It directs the agent that if a command fails due to a sandbox block (runtime_access_denied), it should "re-run it with escalated permissions". This encourages the agent to seek higher privileges (such as sudo or administrative access) to override environment constraints.
[COMMAND_EXECUTION]: The skill provides logic for the agent to select and execute various command-line tools (orca, orca-dev, orca-ide) based on the environment. It instructs the agent to run these tools with various subcommands to manage runtimes, VMs, and containers.
[EXTERNAL_DOWNLOADS]: The skill uses a "discovery stub" pattern where it fetches additional instructions at runtime. The command ORCA skills get orca-per-workspace-env is used to load a "version-matched guide" from an external source, introducing dynamic content into the agent's execution context.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by ingesting external data that influences agent behavior.
Ingestion points: The ORCA skills get command in SKILL.md loads external content into the session.
Boundary markers: None; the skill does not specify delimiters or instructions to ignore embedded commands within the fetched guide.
Capability inventory: The skill has extensive command execution capabilities through the orca CLI tools, as documented in the executable resolution section of SKILL.md.
Sanitization: No evidence of sanitization or validation of the content returned by the external skills get command is provided.

orca-per-workspace-env