From quality
Audit a CLI tool against the agent-friendliness checklist from Zbigniew Sobiecki's "Building Agent-Friendly CLIs". Use when the user says "quality-cli-agent-friendly-audit", "cli-agent-friendly-audit", "audit this cli", "is my cli agent-friendly", or asks to check a CLI for agent ergonomics.
How this skill is triggered — by the user, by Claude, or both
Slash command
/quality:cli-agent-friendly-auditThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Walk through a checklist of properties that make a CLI smooth for AI agents to use. Based on Zbigniew Sobiecki's "Building Agent-Friendly CLIs" (https://zbigniew.me/writing/building-agent-friendly-clis/).
Walk through a checklist of properties that make a CLI smooth for AI agents to use. Based on Zbigniew Sobiecki's "Building Agent-Friendly CLIs" (https://zbigniew.me/writing/building-agent-friendly-clis/).
<cli> --help, <cli> <subcommand> --help, try --json, try a deliberately wrong flag, try a typo, inspect exit codes with echo $?. Don't run anything destructive.Keep the audit terse. One line per item is usually enough. Don't lecture — the user knows what good looks like.
--help and -h both work at every level (root, subcommand, sub-subcommand).--project vs --projectId vs --project-name is a fail).search, get <id>, whoami, recent, or equivalent.whoami (or equivalent) shows which credential is actually in use.--json flag available on every read/list/mutation command.--help or skills).<CLI>_OUTPUT_MODE=json) so agents don't have to remember the flag.--json id,title,status) — bonus.meta.nextCursor or meta.truncated).--all flag drains every page.--json output on stdout is pure JSON — no "Loading..." line, no spinner, no warnings prefixed.NO_COLOR=1 and CI=true.--no-color flag exists for forced override.T-7K9-C) layered over UUIDs.--file <path> or @path prefix) — no shell-quoting hell for multi-line content.0 = success, non-zero = failure. Always.128 + n convention.--help.CI=true, no terminal width).--yes, --force, --non-interactive) — never block on a prompt.Render the audit as a markdown table with three columns: Item, Status, Note. Group by section. At the end add:
## Top fixes (ranked by agent-impact)
1. ...
2. ...
3. ...
Bias the ranking toward fixes that compound across thousands of invocations — stream hygiene, JSON purity, exit code taxonomy, and mutations returning enough info usually beat cosmetic polish.
npx claudepluginhub zcaceres/skills --plugin qualityFetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.
Provides a checklist for code reviews covering functionality, security, performance, maintainability, tests, and quality. Use for pull requests, audits, team standards, and developer training.