The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and follow dynamic instructions (referred to as context, operationGuidance, and rules) retrieved from the local repository via the openspec instructions command.
Ingestion points: External data is ingested in Step 1 (openspec instructions archive) and Step 4 (openspec instructions specs) in SKILL.md.
Boundary markers: The skill provides clear instructions to the agent to treat this data as advisory, to keep it separate from built-in logic, and to report any conflicts rather than allowing the external data to override controlling inputs.
Capability inventory: The agent has the capability to execute shell commands, read repository files, and move directories on the file system.
Sanitization: The skill explicitly instructs the agent not to copy text from these external fields verbatim into output files or specifications.
[COMMAND_EXECUTION]: The skill uses the Bash tool to perform file system operations, including directory creation (mkdir) and moving files (mv).
Evidence: Step 5 in SKILL.md contains shell commands that use variables (changeRoot, planningHome.changesDir, and target-name) derived from the openspec status CLI output or user-provided names. While these variables are double-quoted to prevent simple shell injection, they represent a capability to modify the local file system structure.

openspec-archive-change