提示词库

Coding & Development:Agent 系统提示词。

awesome-prompts 里维护的角色提示词和系统提示词,按领域分组。复制到 Claude Code、Codex 或任意对话模型里就能用。

103 Coding & Development 提示词

A2A Agent Protocol Architect EN

Architect A2A-compliant agent-to-agent systems: AgentCard discovery, Task lifecycle, Message/Part/Artifact contracts, JSON-RPC/gRPC/HTTP bindings, async streaming, OAuth/mTLS security, idempotency, versioning; based on the A2A open protocol (Google → Linux Foundation, v1.0 2026, 22k+ stars, Apache-2.0)

ADHD Parallel Ideation Skill EN

Parallel divergent ideation for coding agents: spawns N isolated branches under cognitive frames (hardware/regulator/biology/speedrunner/etc), scores/clusters/prunes traps, deepens survivors; mechanical generator/critic split with zero shared context during divergence; for architecture, naming, API design, and fuzzy-debugging decisions; based on UditAkhourii/adhd (May 2026, 717+ stars, The New Stack featured, preprint)

Agent Cooperation Designer EN

Prompt for designing cooperative multi-agent systems: shared objective, local roles, disagreement rules, anti-herding controls, evaluation signals (2026)

Agent Cost Observability Architect EN

End-to-end cost observability and budget-governance system for AI coding agents: multi-provider token telemetry, real-time TUI/menubar dashboards, per-project budget envelopes, cost-anomaly detection, optimization recommendation loops, forecast-and-actual tracking; based on getagentseal/codeburn (Apr 2026, 7.2k+ stars)

Agent Harness Designer EN

System prompt for designing reliable agent runtimes: tool minimization, approval gates, memory/compaction, rollback, observability, evals; derived from OpenAI/Anthropic harness guidance (2026)

Agent Harness Performance Engineer EN

Cross-harness agent harness optimization: token economics, memory persistence hooks, continuous learning via instinct extraction, verification loops, parallelization, security scanning; based on affaan-m/everything-claude-code (Jan 2026, 182k+ stars)

Agent Protocol Advisor EN

Prompt for choosing MCP vs A2A vs simpler transports: protocol mapping, trust boundaries, ownership, retries, migration plan; derived from Google's 2026 protocol guide

Agent Skill Designer EN

Prompt for packaging reusable agent skills: narrow scope, tool-aware workflow, safety rules, verification checklist, `SKILL.md` draft output; derived from Anthropic/Google skill guidance (2026)

Agent State Hygiene Architect EN

Local-agent state maintenance architect: inspect-before-mutate discipline, report-first workflow, archive-don't-delete policy, handoff-doc continuity, session metadata bloat detection, stale worktree pruning, log rotation, and config hygiene; based on vibeforge1111/keep-codex-fast (May 2026, 1.2k+ stars)

Agent Virtual Filesystem Architect EN

Unified virtual-filesystem layer for AI agents: mount topology, resource adapters, bash-tool surface, two-layer cache, snapshots/cloning, framework integration; based on strukto-ai/mirage (May 2026, 2149 stars)

Agent-First Language Architect EN

Programming-language designer that treats agents as primary users: small regular surface, deep standard library, deterministic structured tooling, and explicit syntax; based on vercel-labs/zerolang (May 2026, 3.6k+ stars)

Agent-Native CLI Designer EN

Agent-native CLI architect for GUI software: 7-phase SOP to wrap any GUI app into a stateful, agent-usable CLI with REPL + subcommand modes, backend integration, test planning, and SKILL.md generation; based on HKUDS/CLI-Anything (Mar 2026, 34k+ stars)

Agentic Code Reasoner EN

Prompt for evidence-backed code reasoning: semi-formal reasoning chain, competing hypotheses, verification-first conclusions for complex code understanding (2026)

Agentic Coder EN

Plan-first coding agent: security checklist, test discipline, PR summary format (2025)

Agentic HTML Publisher EN

Local-first, ship-ready HTML publisher: turns Markdown/CSV/JSON/notes into single-file HTML via 75 skill templates across 9 surfaces (magazine, deck, poster, social cards, prototype, data report, Hyperframes); juice-inlined CSS for WeChat, 2× PNG for X, standalone .html download; anti-AI-slop design discipline with locked palettes, CJK font stacks, and 8 px baseline grid; based on nexu-io/html-anything (May 2026, 4.5k+ stars)

AGENTS.md Author EN

Authoring prompt for the AGENTS.md open standard: concise repo-root file telling cross-vendor coding agents (Codex CLI, Cursor, Aider, Gemini CLI, Jules, Factory, RooCode; Claude Code via CLAUDE.md) how to set up, build, test, and commit safely; recommended section order, extract-don't-invent commands, monorepo nested-file resolution, ≤200-line discipline, anti-patterns, provenance + questions output; based on the official agents.md spec, OpenAI's Aug 2025 introduction, and Agentic AI Foundation / Linux Foundation 2026 stewardship

Andrej Karpathy Coding Guidelines EN

Concise behavioral guardrails against common LLM coding mistakes: think before coding, simplicity first, surgical changes only, goal-driven verification; derived from Andrej Karpathy's observations on LLM coding pitfalls (Jan 2026)

Anti-AI-Slop Design Architect EN

Structural-variety-first design skill: refuses LLM-default rhythms, enforces 69-gate slop test, locked-token discipline, honest-copy rule, pre-emit 6-axis self-critique, and four verbs (default/audit/redesign/study); based on Nutlope/hallmark (Apr 2026, 2.4k+ stars)

AOS CE Agent Operating System Architect EN

Architect for Unicity AOS Community Edition: capsules, Astrid Runtime, Forge workbench, meta-harness loops, MCP bridge, and least-privilege capability design; based on unicity-aos/aos-ce (Rust, 6.5k+ stars, July 2026)

API Integration Architect EN

Integration architect: pattern selection, auth, retry/backoff, idempotency, observability for reliable system-to-system integrations (2026)

Auditable Enterprise LLM Agent Harness Architect EN

Reconstruct prompt-heavy enterprise LLM prototypes into traceable, auditable, code-owned systems: source-to-claim pipeline, code-owned contracts, seven validation dimensions, replaceable composition boundary, insight-first answer structure; based on "From Prompts to Contracts: Harness Engineering for Auditable Enterprise LLM Agents" (arXiv 2607.08028, July 2026)

Autonomous Permission Classifier Architect EN

Design model-based permission classifiers for coding agents: prompt-injection probe, reasoning-blind transcript classifier with two-stage filter, block/allow templates, deny-and-continue semantics, and recursive subagent handoff gates; based on Anthropic's "How we built Claude Code auto mode" (March 2026)

Autonomous Software Factory Orchestrator EN

Chat-driven autonomous development orchestrator: human sets direction via lightweight messages, self-coordinating claws execute planning/build/test/review/push loops; notification routing (git/tmux/GitHub/lifecycle) kept strictly outside agent context windows; based on ultraworkers/claw-code (Mar 2026, 191k+ stars)

Browser Harness Designer EN

Self-healing browser harness architect: direct CDP websocket, thin editable runtime, agent-generated helper layer, domain/interaction skill separation; based on browser-use/browser-harness (Apr 2026, 12k+ stars)

Classic Software Engineering Canon EN

Multi-book binding ruleset for AI coding agents: Clean Code (readability, naming, functions, side effects), Clean Architecture (dependency direction, boundaries, adapters), Domain-Driven Design (bounded contexts, aggregates, ubiquitous language), Designing Data-Intensive Applications (consistency, durability, replication, schema evolution); unified review checklist; based on ciembor/agent-rules-books (Apr 2026, 1.4k+ stars)

Claude Artifacts EN

System prompt for generating rich Claude Artifacts (UI, interactive apps, code)

Claude Code Loops Operator EN

Turn/goal/time/proactive loop operator for Claude Code: choose the right primitive (`/goal` · `/loop` · `/schedule`), encode verification skills, manage tokens, and design routines that run while you sleep; based on Anthropic's official "Loop engineering: Getting started with loops" guide (July 2026)

Claude Code Sub-Agent Designer EN

Designer prompt for Anthropic's Claude Code sub-agents: when to use sub-agent vs skill vs inline, kebab-case naming, routing description authoring, least-privilege tool allowlists, isolated context discipline, output-contract lock-in, routing stress test; based on Anthropic's Claude Code Sub-Agents docs (Feb 2026) and wshobson/agents + VoltAgent/awesome-claude-code-subagents (2026)

Cline Prompt Architect EN

Cline-optimized prompt engineer: four-element task prompts (goal/context/constraints/done-when), Plan/Act mode discipline, `.clinerules` authoring, MCP server and plugin preferences, multi-agent team scoping, and headless CI/CD conventions; based on cline/cline (Apache-2.0, 64k+ stars, 2026)

Code Reviewer EN

Security-focused code reviewer: OWASP Top 10, severity grading, fix examples (2026)

Codebase Knowledge Graph Architect EN

Transform code, SQL schemas, infrastructure definitions, docs, and multimodal assets into a structured, queryable knowledge graph: AST-level entity extraction, God-node identification, surprising cross-module connections, design-rationale mining, architectural tension detection, and confidence-tagged edges (EXTRACTED / INFERRED / AMBIGUOUS); outputs GRAPH_REPORT.md, graph.json, and optional interactive visualization; supports incremental delta updates on commits; based on safishamsi/graphify (Apr 2026, 44k+ stars)

Codebase Memory MCP Architect EN

MCP-native code-intelligence architect for DeusData codebase-memory-mcp: index repos into a persistent knowledge graph (158 languages, Hybrid LSP, <1ms structural queries), map agent questions to the 15 MCP tools, design indexing/watch/artifact policies, and enforce query plans that replace file-by-file exploration; based on DeusData/codebase-memory-mcp (MIT, 37k+ stars, Feb 2026)

Coding Agent System Prompt EN

Production-grade system prompt for CLI coding agents: identity, permission model, task execution discipline, code style constraints, risk-aware action, tool usage protocol, output efficiency; independently authored from patterns observed in Claude Code (Apr 2026)

Computer Use Operator EN

System prompt for browser/desktop agents: observe → act → verify loops, least privilege, confirmation gates, phishing/prompt-injection resistance; derived from OpenAI's 2026 computer-use guidance

Database Schema Designer EN

DB architect: entity modeling, normalization (1NF, 3NF), index strategy, PostgreSQL DDL with migration notes (2026)

Debugging Agent EN

Systematic bug hunter: reproduce → observe → hypothesize → test → localize → fix; works for any language (2026)

Design System Spec Architect EN

Prompt for authoring DESIGN.md design-system specifications: machine-readable YAML tokens + human-readable rationale, component definitions, state variants, and WCAG-safe palettes; derived from Google Labs' 2026 design.md specification (2026)

Disciplined Diagnostician EN

Disciplined diagnosis loop for hard bugs and performance regressions: feedback-loop construction, falsifiable hypotheses, instrumented probes, correct regression-test seams, cleanup protocol; based on mattpocock/skills (Feb 2026)

Doubt-Driven Development Architect EN

Fresh-context adversarial review for non-trivial decisions: CLAIM → EXTRACT → DOUBT → RECONCILE → STOP cycle; isolates artifact + contract, forbids passing the claim to the reviewer, bounds doubt theater, offers cross-model escalation; based on addyosmani/agent-skills (2026, 54.7k+ stars)

Emotion-Aware Engineering Partner EN

Senior coding partner grounded in Anthropic's 2026 emotion-vectors research: incremental delivery, honest uncertainty calibration, collaborative pushback, debugging transparency (2026)

Fable Method Agent Loop Architect EN

Think / act / prove agent loop: classify the ask, define done with named verification, gather primary-source evidence in parallel, commit to one recommendation, act surgically, verify by observation, report outcome-first; includes domain adapters, triviality/fit/intent/recall/authorization gates, twin-check, and artifact gate; based on Sahir619/fable-method (MIT, 1.9k+ stars, July 2026)

Formal Theorem Proving Architect EN

Blueprint-driven Lean 4 prover: dependency-graph decomposition, parallel lemma proving, compiler-feedback refinement loops; 99.2% pass@1 on MiniF2F-test, 75.6% on PutnamBench; based on Goedel-Architect (arXiv 2606.06468, June 2026)

Frontend Developer EN

React/Vue/Angular expert: component architecture, Core Web Vitals, WCAG 2.1, responsive design, TypeScript, performance budgets (2026)

Frontend Taste Engineer EN

Senior UI/UX engineer that overrides default LLM biases toward generic UI: metric-based design rules (variance/density/motion dials), anti-slop guardrails, CSS hardware acceleration, spring physics, liquid-glass refraction, and premium interaction states; based on Leonxlnx/taste-skill (Apr 2026, 17.5k+ stars)

Gemini CLI Prompt Architect EN

Gemini-CLI-optimized prompt engineer: four-element task prompts (goal/context/constraints/done-when), GEMINI.md discipline, built-in tool preferences (search/file/shell/fetch), MCP @-server mentions, multimodal inputs, and anti-patterns; based on google-gemini/gemini-cli (Apache-2.0, 105k+ stars, 2026)

Generative UI Architect EN

Component-first, design-system-native UI generation: states, tokens, accessibility, responsive layouts, typed code output (2026)

Goose AI Engineering Agent Operator EN

Vendor-neutral open-source AI engineering agent operator: MCP-native extension discipline, plan-then-execute loops, multi-provider awareness, least-privilege permission model; based on block/goose → aaif-goose/goose under the Linux Foundation Agentic AI Foundation (Apache-2.0, ~50k stars, June 2026)

Grok Build Prompt Architect EN

Grok-Build-optimized prompt engineer: four-element task prompts (goal/context/constraints/done-when), AGENTS.md / CLAUDE.md project-rule discipline, `.grok/skills/` authoring, TUI slash commands (`/compact`, `/fork`, `/rewind`), headless `grok -p` / ACP `grok agent stdio` scoping, MCP-aware tool preferences, permission rules, and sandbox profiles; based on xai-org/grok-build (Apache-2.0, 18k+ stars, July 2026)

HTML PPT Studio Designer EN

Professional static HTML presentation architect: 36 themes, 15 full-deck templates, 31 layouts, 47 animations (27 CSS + 20 canvas FX), true presenter mode with pixel-perfect previews + speaker script + timer; token-based design system, keyboard runtime, no build step; based on lewislulu/html-ppt-skill (Apr 2026, 4676 stars)

HTML-Native Design Orchestrator EN

Single-sentence-to-ship design skill: interactive prototypes, HTML decks, motion design (MP4/GIF), infographics, and 5-dimension expert critique; enforces Core Asset Protocol (logo → product shots → UI → color → font), Junior Designer workflow, anti-AI-slop rules, and 5-schools×20-philosophies design direction advisor; based on alchaincyf/huashu-design (Apr 2026, 14k+ stars)

Improve Audit Planner EN

Codebase audit → self-contained plans → cheap-executor dispatch: nine-dimension audit with `file:line` evidence, machine-checkable verification gates, isolated worktree execution, and backlog reconciliation; based on shadcn/improve (MIT, 8.6k+ stars, June 2026)

Jetpack Compose Architect EN

Production-grade Jetpack Compose code architect: state authoring/hoisting/holder patterns, recomposition performance, stability diagnostics, deferred reads, side-effect lifecycle, Kotlin Flow state/event modeling, accessibility and Material 3 compliance; based on chrisbanes/skills (May 2026, 660 stars)

Kimi Code Prompt Architect EN

Kimi-Code-CLI-optimized prompt engineer: four-element task prompts (goal/context/constraints/done-when), `coder`/`explore`/`plan` subagent selection, `/goal` judge-verified stop conditions, AI-native `/mcp-config`, SKILL.md authoring, lifecycle hooks, video/multimodal input, and KIMI.md/AGENTS.md project-rule discipline; based on MoonshotAI/kimi-code (MIT, 6.2k+ stars, May 2026)

Launch Your Agent Architect EN

Founder copilot for launching Claude Managed Agents (CMA): interview a founder, scope the smallest v0, launch in their own Anthropic account, grade against a binary rubric, iterate, and schedule deployments; based on anthropics/launch-your-agent (Apache-2.0, June 2026)

Loop Engineering Architect EN

Design external loop specifications that let coding agents run without step-by-step prompting: trigger, goal, five-level verification ladder, architecture, stopping rule, durable memory; based on "Stop Hand-Holding Your Coding Agent" (arXiv 2607.00038, July 2026)

Loop Engineering Patterns Operator EN

Practical loop pattern operator for recurring coding-agent tasks: select from 7 production patterns (PR Babysitter, Daily Triage, CI Sweeper, etc.), scaffold with `loop-init`, score Loop Ready with `loop-audit`, and operate the five building blocks + memory across Grok, Claude Code, Codex, and Opencode; based on cobusgreyling/loop-engineering (MIT, 9.7k+ stars, June 2026)

Magazine Web Deck Designer EN

Single-file HTML horizontal-swipe deck architect: two locked visual styles (Editorial Magazine × Electric Ink vs Swiss Internationalism), WebGL hero backgrounds, 10, 22 registered layout skeletons, locked theme presets, Motion One choreography, typography-first discipline; based on op7418/guizang-ppt-skill (Apr 2026, 8590 stars)

Managed Agent Architect EN

Prompt for designing long-running managed-agent systems: brain/hands split, worker contracts, checkpoints, permission scoping, recovery; derived from Anthropic/OpenAI 2026 harness guidance

MiMo Code Prompt Architect EN

MiMo-Code-optimized prompt engineer: four-element task prompts (goal/context/constraints/done-when), build/plan/compose agent selection, persistent SQLite FTS5 memory (MEMORY.md / checkpoint.md / tasks), `/goal` judge-verified stop conditions, compose-mode specs-driven workflows, deterministic JS workflows, and `.mimocode/skills/` authoring; based on XiaomiMiMo/MiMo-Code (MIT, 12k+ stars, June 2026)

Mobile App Builder EN

Native iOS (Swift/SwiftUI) + Android (Kotlin/Jetpack Compose) + cross-platform (React Native/Flutter): offline-first, biometric auth, push notifications, app store deployment (2026)