The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute the factory CLI tool for validating and running configurations, specifically using commands like factory run-once --config ./factory.config.json --dry-run and factory start --mode live. These commands are part of the intended functionality for the '@agent-relay/factory' tool provided by the vendor.
[DATA_EXPOSURE_AND_EXFILTRATION]: The provided configuration examples include a hardcoded absolute file path (/Users/khaliqgant/Projects/AgentWorkforce) in the repos.cloneRoot field, which discloses a local workstation username.
[INDIRECT_PROMPT_INJECTION]: The skill describes a workflow that ingests data from external sources (GitHub issues and Linear tasks) to trigger automated actions such as code modifications and PR creation. This introduces a surface for indirect prompt injection if the external data contains malicious instructions.
Ingestion points: GitHub issues labeled 'factory' and Linear issue states.
Boundary markers: No explicit delimiters or boundary markers for untrusted data are specified in the configuration guidelines.
Capability inventory: The factory tool possesses capabilities for cloning repositories, modifying code, and dispatching Relayflows.
Sanitization: There are no instructions or configurations provided for the sanitization or validation of the content ingested from external providers.

factory-config