The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates shared memory by reading and writing Markdown files in a workspace's Git directory. This creates a surface where an agent might ingest instructions or malicious content written by other agents or users into its context during subsequent sessions.
Ingestion points: Markdown files located in the workspace Git folder.
Boundary markers: The instructions do not specify any delimiters or markers to distinguish shared memory content from the agent's core instructions.
Capability inventory: The skill utilizes the leverage CLI tool for session management and standard agent capabilities for reading/writing files.
Sanitization: No explicit sanitization or validation of the shared memory content is described before it is read by the agent.
[COMMAND_EXECUTION]: The skill provides instructions for executing vendor-specific CLI tools to manage user identity and session memory.
Evidence: leverage whoami --json, leverage sessions new.
[EXTERNAL_DOWNLOADS]: The skill suggests installing additional functionality from the author's own repository using a package runner.
Evidence: npx skills add leverage-computer/skills --skill leverage-cloud-agents.

team-agent-memory