owenob1/prompt-improver / skills/prompt-improver
prompt-improver
VerifiedTransform vague prompts into precise, verifiable structured XML prompts that coding agents execute reliably. Modes: execute (default — generate then run) and plan (generate XML for review first). Use when the user says improve prompt, make this work better, prompt engineer, structure a request, plan a complex change before coding, or when a rough request needs verification criteria before execution. Do not use when the input is already a well-structured XML prompt or detailed implementation spec — skip generation and execute directly.
config/settings.example.json
View on GitHub{
"version": 1,
"//": "Copy to ~/.config/prompt-improver/settings.json or .prompt-improver/settings.json. Keys starting with // are comments only (ignored). Full guides: docs/CUSTOM-BACKENDS.md, references/models-supported.md. Shipped tables: config/runtime-defaults.json.",
"//backend": "auto | claude | grok | gemini | codex | agy | copilot | cursor | opencode | cline | qwen | droid | amp | kimi | kiro — Env: PROMPT_IMPROVER_BACKEND",
"backend": "auto",
"//model": "Force one model (null = default_models[backend]). Env: PROMPT_IMPROVER_MODEL. Per-prompt model:<id> wins.",
"model": null,
"//default_models": "Generator model per CLI when model is null. null = leave the CLI's own default (multi-provider CLIs use their own model ids). claude uses the `opus` alias, which tracks the newest Opus.",
"default_models": {
"claude": "opus",
"grok": "grok-4.7",
"gemini": "gemini-3.8-flash",
"codex": "gpt-6-sol",
"agy": null,
"copilot": null,
"cursor": null,
"opencode": null,
"cline": null,
"qwen": null,
"droid": null,
"amp": null,
"kimi": null,
"kiro": null
},
"//max_tokens": "Soft output cap (best-effort). Env: PROMPT_IMPROVER_MAX_TOKENS",
"max_tokens": 12000,
"//enable_research": "false = omit <research> in generated prompts",
"enable_research": true,
"//enable_thinking": "false = omit <approach> blocks unless required",
"enable_thinking": true,
"//headless_only": "true = host must not in-session rewrite when headless fails",
"headless_only": true,
"//allow_web_search": "false = no web search in generated executor instructions",
"allow_web_search": true,
"//allow_code_execution_in_generation": "false = generator cannot run gather-context.sh or shell",
"allow_code_execution_in_generation": false,
"//skip_validate": "true = skip validate-prompt.sh after generation",
"skip_validate": false,
"//backend_invocation": "scripts | commands | auto — how built-in CLIs are invoked",
"backend_invocation": "scripts",
"//fallback_strategy": "manual | error",
"fallback_strategy": "manual",
"//preferred_backends": "Order to try OTHER generators after the primary fails — not the default pick. kimi is left out because its -p mode auto-approves tools.",
"preferred_backends": [
"claude",
"codex",
"grok",
"gemini",
"agy",
"copilot",
"cursor",
"opencode",
"cline",
"qwen",
"droid",
"amp",
"kiro"
],
"//custom_command": "Escape hatch: prompt on stdin, improved text on stdout. Bypasses backends. Env: PROMPT_IMPROVER_CUSTOM_COMMAND",
"custom_command": null,
"//model_aliases": "Override/extend alias→canonical map (merged with runtime-defaults.json). Example:",
"//model_aliases_example": {
"my-alias": "sonnet"
},
"//model_fallback_chains": "Replace entire chain table when set. Example:",
"//model_fallback_chains_example": [
{
"patterns": [
"sonnet",
"*sonnet*"
],
"chain": [
"$primary",
"haiku"
]
}
],
"//backend_commands": "Override CLI templates (used when backend_invocation is commands, or auto with an override here). Placeholders: {prompt_file} {model} {model_args} {max_tokens}; {model} is shell-quoted. Keep generators read-only.",
"//backend_commands_example": {
"grok": "grok -p \"$(cat \"{prompt_file}\")\" --output-format plain --always-approve --no-subagents --no-plan --disable-web-search --max-turns 3 {model_args} </dev/null"
},
"//supported_backends": "Allowlist for host-matched headless generation",
"//limit_detection": "Regex strings: account_patterns, retry_patterns, xml_markers",
"notes": "Default pick: host CLI + default_models (not first-on-PATH). Override any runtime-defaults.json key in this file. model: routes by family. No host generator → HOST_BOUNCE:NO_HEADLESS.\nGeneration materials + context_mode: see generation object / runtime-defaults.json.",
"generation": {
"context_mode": "deterministic",
"include_xml_template": true,
"include_principles": true,
"include_chaining": true,
"include_examples": true,
"include_system_prompt": true,
"system_prompt_path": "assets/generation-agent-prompt.md",
"xml_template_path": "references/xml-template.md",
"principles_path": "references/prompting-principles.md",
"chaining_path": "references/prompt-chaining.md",
"examples_path": "examples/before-after.md",
"extra_reference_paths": [],
"output_instructions": "Output ONLY the final improved XML prompt. No explanation, no code fences, no commentary, no execution of the request.",
"require_xml_output": true,
"forbid_agent_codebase_search": true,
"backend_timeout_secs": 300,
"grok_timeout_secs": 180,
"grok_max_turns": 3
},
"//generation": "Customise generation materials and deterministic context. Full defaults: config/runtime-defaults.json → generation"
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Unverified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |