The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands including openspec, mkdir, and mv to manage workflow artifacts and the file system. It uses double-quoting for variables like <name>, <changeRoot>, and <target-name> to mitigate basic command injection risks.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external data from the openspec instructions command, including context, operationGuidance, and rules fields.
Ingestion points: JSON outputs from openspec instructions archive and openspec instructions specs are loaded into the agent's context.
Boundary markers: The skill contains detailed instructions to treat this external data as advisory, keep it separate from core logic, and never copy it verbatim into output files.
Capability inventory: The skill has the ability to move directories (mv), create folders (mkdir), and perform intelligent merges (spec synchronization) which involves writing to main specification files.
Sanitization: It includes logic to detect and report conflicts between external guidance and its internal controlling inputs, prioritizing the latter to prevent behavior overrides.
[DYNAMIC_EXECUTION]: The skill dynamically invokes a sub-workflow, openspec-sync-specs, at runtime to handle the complex task of merging delta specifications into main specifications.

openspec-archive-change