The Agent Skills Directory

[PERSISTENCE]: The skill utilizes the CronCreate tool to schedule a weekly 'Coverage Drift Monitor'. This creates a persistent background job that executes an agent prompt on a weekly schedule to detect regressions in test coverage or failures in existing tests.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data including user-supplied scope arguments and project source code, which are interpolated into the prompts for specialized subagents (e.g., ork:test-generator).
Ingestion points: The SCOPE variable (derived from $ARGUMENTS) and source code contents (identified during discovery and risk analysis) are passed into agent prompts.
Boundary markers: The skill does not employ explicit boundary delimiters or 'ignore embedded instructions' warnings when passing this data to subagents.
Capability inventory: The skill and its subagents have access to powerful tools including Bash, Write, Workflow, Agent, SendMessage, and PushNotification.
Sanitization: No sanitization, escaping, or filtering of the external source code or user input is performed before interpolation into prompts.
[COMMAND_EXECUTION]: The skill performs extensive shell command execution for framework detection, running existing test suites to establish baselines, and executing generated tests. It also invokes a local Node.js utility script (scripts/check-behaviour-tests.mjs) to perform static analysis on new test files.
[DYNAMIC_EXECUTION]: The skill dynamically generates test files and executes them. It also employs a complex 'heal loop' implemented as a JavaScript workflow (workflows/heal-loop.js) which manages the iterative diagnosis and repair of test failures by spawning additional agents and verifying their edits.
[DYNAMIC_CONTEXT_INJECTION]: The skill's frontmatter includes invocation_hooks that execute shell commands (command -v vitest, etc.) at load time to verify the environment. While these specific commands are benign checks, they represent execution at the point of skill invocation.

cover