The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data provided by the openspec instructions archive and openspec instructions specs commands, which can supply a context and operationGuidance to influence the agent's behavior.
Ingestion points: SKILL.md (Steps 1 and 4) executes shell commands and parses their JSON output to obtain behavioral guidance.
Boundary markers: The instructions explicitly direct the agent to keep external guidance separate from built-in steps, report conflicts, and never allow advisory inputs to override controlling inputs or command contracts.
Capability inventory: The skill uses the Bash(openspec:*) tool to run CLI commands and perform file system operations like mkdir and mv.
Sanitization: The skill contains logic to handle non-zero exit codes or invalid JSON from these commands by proceeding without context rather than failing or executing arbitrary code.
[COMMAND_EXECUTION]: The skill constructs and executes shell commands for file management based on paths retrieved from the CLI tool's status output.
Evidence: Step 5 uses mkdir -p and mv to relocate the changeRoot directory to an archive path, both of which are variables derived from the execution of openspec status --json.

openspec-archive-change