The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources that could contain malicious instructions.
Ingestion points: The agent reads JSON data from openspec instructions archive and openspec instructions specs (Step 1 and 4), and parses user-controlled tasks.md files (Step 3).
Boundary markers: The skill includes explicit defensive instructions, such as treating external fields as advisory, requiring conflicts to be reported rather than automatically followed, and forbidding the verbatim copying of external instruction text into specifications or archive summaries.
Capability inventory: The skill utilizes Bash for command execution, including file moves (mv), directory creation (mkdir), and openspec CLI operations.
Sanitization: The workflow requires validation of CLI exit statuses and JSON formats, and instructs the agent to preserve its built-in behavioral contracts over external guidance.
[COMMAND_EXECUTION]: The skill performs shell-level operations to manage the project state.
Evidence: The instructions use mkdir -p and mv to manage the archive directory structure and relocation of the changeRoot. It also heavily relies on the openspec CLI tool for status checks and workflow orchestration.

openspec-archive-change