The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to configure the agent with official MCP servers from a well-known service.
Evidence: Fetches and installs @anthropic/memory-mcp-server, @anthropic/scheduled-tasks-mcp-server, and @anthropic/computer-use-mcp-server using npx for configuration in ~/.claude.json.
[COMMAND_EXECUTION]: The skill uses programmatic commands and automated scheduling tools to create a persistent agent loop.
Evidence: Employs claude -p for programmatic execution and mcp__scheduled-tasks__create_scheduled_task to establish recurring tasks (crons) that execute arbitrary prompts on a schedule.
[PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by design through its autonomous data processing workflows.
Ingestion points: Ingests untrusted data from external sources including GitHub pull requests, Exa search results, and calendar invite details as part of its automated background tasks.
Boundary markers: The example prompts provided for the autonomous reviewer and research agent lack explicit delimiters or instructions to ignore embedded commands within the fetched external content.
Capability inventory: The agent is granted extensive capabilities including local file system writes to memory stores, command-line execution via claude -p, and full browser/desktop control via the computer-use MCP server.
Sanitization: There is no mention of sanitization, filtering, or validation of the external content before it is interpolated into the agent's context for decision-making.

autonomous-agent-harness