The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the openspec CLI tool to retrieve project context at the start of a session.
Evidence: Execution of openspec list --json to identify active changes and schemas.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect injection as it ingests untrusted content from project-level markdown files.
Ingestion points: Local project files including proposal.md, design.md, tasks.md, and GLOSSARY.md (specified in SKILL.md).
Boundary markers: Absent; the skill reads these files and incorporates their content directly into the conversation context.
Capability inventory: Includes execution of the openspec CLI and several user-interaction tools (request_user_input, question, ask_question).
Sanitization: Absent; the skill relies on the underlying LLM's guardrails when processing file content.

openspec-grill