AI Agent & Automation Workflows

Create a new Claude Code agent file (review or team type) following the official sub-agent schema and token-efficiency budgets. Use when the user wants to add a new review agent, detect a new category of code issue, create a team agent persona, or says things like "add an agent for X", "create a reviewer for Y", "new team agent for Z". Also use when given a URL to a coding standard that should become a review agent.

Conventions, anti-patterns, and meta-patterns for writing skills (and the shared agent/skill philosophy). Use when creating or editing a SKILL.md file, or when reviewing the agent-vs-skill separation. For the procedural workflow that generates a new agent file, use the agent-create skill (invoked by /agent-add).

Decide which agents and skills to load for a given task. Use at the start of every task to select the minimum viable context load, calculate the token budget, and stay below the 40% utilization ceiling.

Validate Gherkin feature files for structural quality, determinism, and implementation independence, then verify each scenario has matching test automation. Use this skill whenever reviewing test files, feature files, or BDD scenarios — including during /code-review when .feature files or step definition files appear in the changeset. Also use when a user asks to "check my feature files", "validate my Gherkin", "are my scenarios testable", or "do my feature files have tests".

Approval gates, intervention commands, and transparency requirements. Use to classify any agent action as autonomous/notify/approve, respond to override/pause/stop commands, or structure a plan review before the implementation phase begins.

Safely modify code that lacks tests. Use whenever tasked with changing code without test coverage — apply characterization tests and dependency-breaking techniques before making any behavioral changes.

Log task completion data to .claude/metrics/. Use at the end of every task to record tokens, cost, agents used, rework cycles, and hallucination events. Also use for periodic reporting to identify efficiency and quality trends.

Detect business logic reimplemented in multiple architectural layers. Builds a persistent computation-register.json by annotating non-trivial computation functions with structured semantic descriptions, then clusters entries to surface duplicate domain concepts. Runs in full-scan mode on first use, incremental (git-diff-based) mode on subsequent runs. Use when the user wants to find logical duplication that linters and diff-scoped review agents miss — the same domain calculation independently reimplemented across layers.