The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions explicitly direct the agent to ingest and follow instructions from external data sources via the openspec instructions archive command.
Ingestion points: Data is ingested from the JSON output of openspec instructions archive in SKILL.md (Step 1).
Boundary markers: The skill lacks explicit boundary markers or instructions to the agent to treat the external context or operationGuidance as untrusted data.
Capability inventory: The skill has the capability to execute shell commands (Bash tool), write/sync specifications, and move directories on the file system.
Sanitization: There is no sanitization of the fetched context or guidance; the skill states that context must be treated as a 'required prompt-level input'.
[COMMAND_EXECUTION]: The skill uses shell commands like mkdir -p and mv to manage the archive directory structure. While it uses variables for paths, they are wrapped in double quotes to prevent basic shell injection, though they still rely on the integrity of the paths provided by the openspec CLI output.

openspec-archive-change