Skip to main content

Toolkit

Commands & Workflows · Pitfalls

On this page7 sections

Pitfalls

Assuming Slash-Name Portability

The same command may be absent or mean something different in another product. Port the workflow contract.

Ambiguous Arguments

Natural-language destinations can select the wrong environment or scope. Resolve and echo exact targets before acting.

Hidden Authority Expansion

A command called ship does not imply permission to push, deploy, message customers, or spend. Require confirmation at each material boundary.

Duplicated Command and Skill Logic

Two procedures drift. Make the command a thin adapter to one reviewed workflow.

Session-State Dependence

A workflow that relies on unrecorded prior conversation cannot be reproduced or audited. Declare required inputs.

Success Without Evidence

Require artifact paths, test output, external identifiers, and failures. Narrative confidence is not completion.

Falsification Checklist

  • Does invocation resolve exact inputs and scope?
  • Is the workflow identical across adapters?
  • Can interruption or duplicate invocation cause damage?
  • Are material side effects separately approved?
  • Can another person verify the result without the conversation?

Source register

Commands & Workflows sources

Primary vendor documentation reviewed on the date shown. Links may change after publication.

  1. Extend Claude with skills (opens in a new tab)

    Anthropic · official docs · accessed 2026-08-01

  2. Claude Code commands (opens in a new tab)

    Anthropic · official docs · accessed 2026-08-01

  3. Build skills (opens in a new tab)

    OpenAI · official docs · accessed 2026-08-01

  4. Codex slash commands (opens in a new tab)

    OpenAI · official docs · accessed 2026-08-01

  5. GitHub Copilot CLI command reference (opens in a new tab)

    GitHub · official docs · accessed 2026-08-01