The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local CLI commands via the openspec tool to list, check status, and sync project specifications. It also uses shell commands like mkdir and mv to manage the directory structure of the changes.- [PROMPT_INJECTION]: The skill includes positive guardrails requiring the agent to use the AskUserQuestion tool to let the user select changes, explicitly forbidding the agent from guessing or auto-selecting inputs.- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by reading and processing external data from tasks.md and delta specification files to determine the workflow state. * Ingestion points: reads tasks.md and directory contents in openspec/changes/<name>/specs/. * Boundary markers: None; the skill reads file content directly to identify task completion markers and compare specification text. * Capability inventory: File system manipulation (mkdir, mv), subagent invocation, and local CLI execution (openspec). * Sanitization: No explicit sanitization or filtering of the content retrieved from tasks.md or specification files is performed before the agent evaluates the state.

openspec-archive-change