The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes active conversation history and session data to generate structured summaries that are then written to PR/MR descriptions. This creates a surface where malicious instructions embedded in the processed data could be persisted in the repository.
Ingestion points: Conversation context and session data retrieved via the opencode export command in SKILL.md.
Boundary markers: The output is formatted within a structured JSON block inside a Markdown <details> tag, which helps delineate the AI content from the rest of the PR description.
Capability inventory: The skill possesses the ability to modify remote repository metadata using gh pr edit and glab mr update in SKILL.md.
Sanitization: The instructions provide a list of sensitive data types to exclude, but the skill relies on the agent's adherence to instructions rather than programmatic sanitization or filtering.
[COMMAND_EXECUTION]: The skill relies on shell command execution to interface with source control platforms and session management tools.
Evidence: Uses gh (GitHub CLI) for pr edit and pr view, glab (GitLab CLI) for mr update, and opencode for session data export as defined in SKILL.md.
Context: These operations use standard, well-known developer tools to fulfill the skill's primary function.

session-export