| name | rsh-product |
| description | Shape Restish product direction, feature strategy, and CLI/developer-tool UX. Use when planning, prioritizing, designing, or reviewing a Restish feature before or during implementation; when evaluating whether a request fits Restish; when turning user feedback into a feature proposal; when checking CLI UX, help, errors, output, docs, tests, compatibility, or release readiness. |
Restish Product
Use this skill to keep Restish feature work anchored in user value, CLI ergonomics, and maintainable open-source strategy. Product work is not separate from implementation: it is the thread connecting the user problem, command shape, output contract, tests, docs, and release story.
For source-backed rationale and examples from product/UX references and popular open-source CLI projects, read references/research-notes.md when you need deeper context or are updating this skill.
Core Stance
- Start with the user job, not the implementation hook.
- Prefer a small valuable change over a broad configurable surface.
- Treat docs, help text, errors, output shape, exit codes, and non-TTY behavior as product surface.
- Design for both interactive humans and scripted automation.
- Preserve Restish's role: an API-aware HTTP CLI that stays composable with the shell.
- Avoid building a feature just because it is easy, internally elegant, or possible from an OpenAPI field.
Use This Skill For
- Feature discovery, product shaping, and prioritization.
- CLI command naming, flags, prompts, defaults, help, errors, and output design.
- Comparing alternate product directions for Restish.
- Reviewing a proposed change for user value, scope creep, discoverability, and compatibility.
- Writing design-doc product sections, issue proposals, release notes, or UX test plans.
Do Not Use This Skill For
- General code review after a patch exists. Use
rsh-review.
- Documentation-only execution once the product direction is settled. Use
rsh-docs.
- Approved simplification or dead-code cleanup. Use
rsh-simplify.
- Pure implementation where the behavior, UX, and scope are already explicit.
Workflow
1. Frame The Opportunity
Before designing commands, write the smallest useful product frame: who the user is (first-time user, daily CLI user, API integrator, plugin operator, plugin author, or maintainer), the job they are doing outside Restish, their current workaround, the pain, the intended outcome, and the evidence (issue text, failing example, repeated support question, code/test signal). If evidence is thin, propose the lightest validation instead of inventing a large feature.
Concrete evidence sources for Restish: the v1 issue tracker (rest-sh/restish on GitHub — dozens of open issues and PRs represent real user demand), docs/design/000-restish-v1-baseline.md for v1 behavior, docs/design/031-compatibility-and-migration.md for compatibility promises, and docs/design/037-v2-command-surface-review.md for prior command-surface decisions.
2. Classify The Investment
Classify the work — reliability, usability, feature, compatibility, or platform — so scope and validation match risk. Favor reliability and usability improvements when a new feature would deepen existing confusion. Platform work counts only when it unlocks user-facing reliability, velocity, or simplicity.
3. Shape The Product Bet
For meaningful changes, produce a short product note or design-doc section:
## Product Frame
Problem:
Users need ...
Goals:
- ...
Non-goals:
- ...
Primary workflow:
1. ...
CLI surface:
- Command/flag/config:
- Defaults:
- TTY behavior:
- Non-TTY behavior:
- Output:
- Errors:
Compatibility:
- Existing behavior preserved:
- Intentional break, if any:
Validation:
- Tests:
- Docs/help:
- Manual checks:
Open questions:
- ...
Keep the bet rough, solved, and bounded: enough detail to reduce risk, not so much that implementation has no room to find the right shape.
4. Apply Restish Product Principles
Use these as decision filters:
- API-aware, shell-native: Generated OpenAPI commands should feel richer than raw HTTP, but generic URL requests must remain excellent.
- Human first, machine always: Human output may be friendly in a TTY; machine-readable output must be stable, color-free when piped, and easy to combine with
jq, grep, files, and scripts.
- Examples before reference: Help and docs should lead with real commands and representative output.
- Safe by default: Destructive, auth-sensitive, TLS-sensitive, or remote state-changing flows need clear confirmation, dry-run, or explicit force paths when appropriate.
- No mystery waits: Network, pagination, streaming, plugin, and subprocess flows need responsive feedback or bounded timeouts.
- Progressive disclosure: Common workflows should be simple; advanced auth, content negotiation, output, plugin, and OpenAPI controls should be discoverable without crowding the default path.
- Convention over novelty: Reuse familiar flag names and CLI conventions unless Restish has a strong reason to differ.
- Few knobs, strong defaults: Add configuration only when users need durable variation. Do not add parsed-but-unused or speculative config.
- Compatibility is product: Preserve documented behavior, v1 migration promises, plugin protocol contracts, and scriptability unless the product note calls out an intentional break.
- Docs are part of done: User-visible behavior needs
site/ updates; significant architecture or behavior needs docs/design/ updates.
5. Design The CLI UX
Check every new or changed command, flag, output mode, or prompt:
- The command name matches the user's object and action vocabulary.
-h and --help work at the command and subcommand level.- Top-level and subcommand help include the most common examples first.
- Required input can be supplied non-interactively; prompts are TTY-only.
- Dangerous operations have confirmation,
--dry-run, --force, or equivalent safety where appropriate.
stdout is reserved for primary output; diagnostics and progress go to stderr.- TTY output may use color, tables, summaries, and progress; piped output must be clean and deterministic.
- JSON, YAML, raw, table, streaming, pagination, filtering, and redirect behavior remain compatible with existing output contracts.
- Errors say what failed, why it matters, and what the user can do next.
- Exit codes distinguish success from failure and preserve script expectations.
- Ctrl-C, timeouts, retries, plugin subprocesses, and partial output fail in understandable and recoverable ways.
- Config, environment variables, flags, profiles, and API-generated defaults have clear precedence.
6. Validate Before And After Implementation
Use validation proportional to risk:
- Product validation: Compare against user jobs, known feedback, current docs, and one or two realistic command transcripts.
- Usability review: Run a heuristic pass for status visibility, vocabulary, control/escape, consistency, error prevention, recognition over recall, efficiency, minimal output, error recovery, and help.
- CLI contract tests: Cover TTY vs non-TTY when behavior differs,
stdout/stderr, exit codes, help text, output formats, malformed input, cancellation, auth/config precedence, and shell composition.
- Output regression tests: Use for intentional formatter/help changes; do not accept drift casually.
- Docs checks: Update
site/ for user behavior and docs/design/ for significant subsystem behavior; link guides and reference pages both ways.
- Compatibility checks: Test v1-style workflows, generated OpenAPI command shape, plugin boundaries, and scripting behavior when touched.
Output Expectations
When advising, lead with the recommendation and the reason. Then include:
- Product frame or decision.
- Tradeoffs and rejected alternatives.
- CLI UX implications.
- Validation plan.
- Docs/design-doc impact.
When implementing, leave a visible product trail in the design doc, issue note, tests, docs, help text, or final summary. Do not bury product decisions only in code.
