AI Tools
140 prompts in this category · Page 11 of 12
Continue.dev Config from a Repo Stack Brief
Write a Continue.dev config.yaml from a repo stack brief. No invented models or other-IDE rule files.
Act as a Continue.dev power user writing config.yaml from a repo stack brief. You write Continue.dev config.yaml from a stack brief. You lock model names to Inputs. You do not emit Cursor AGENTS.md, Copilot, Cline, Windsurf, or JetBrains AI files unless Inputs ask. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Continue version or config schema I want: [Version] - Repo languages, package managers, test command: [Stack] - Models I may call (provider plus model id): [Models] - Context providers I allow: [Context] - Rules I want in config (never-list, style): [Rules] - MCP or tools I actually have: [Tools or NONE] - Auth note (env var name only): [Auth] - What other assistant files I must not write: [NotFiles] - Tab autocomplete model or NONE: [Tab] - Words I must not use: [Banned] Generate: 1. Version lock: quote Version. If unknown, write config.yaml for Continue 1.1.x style and mark schema unknown. 2. Honesty ledger: languages, test command, model ids, providers, context providers, tools. Forbidden: models not in Models, files in NotFiles. 3. config.yaml sketch: models list from Models only, tabAutocompleteModel from Tab or omit, contextProviders from Context. 4. Rules block: never-list from Rules. Include Stack test command as the only test command. 5. Tools: MCP only if Tools names them. If NONE, omit. 6. Auth: print only the env var name from Auth. Do not print a token. 7. Refuse other files: list NotFiles you will not emit. 8. Compliance pass: quote Banned words, invented model ids. Cut them. Constraints: - Continue.dev config.yaml only. Do not write Cursor, Copilot, Cline, Windsurf, or JetBrains AI files. - Never invent a model id. - Test command only from Stack. - If Tab is NONE, omit autocomplete rather than guessing a small model. - No emojis.
Cline and Windsurf Repo Rules File from a Stack Brief
Write a Cline or Windsurf repo rules file from a stack brief. Not Cursor AGENTS.md. No invented CLI commands or versions.
Act as a staff engineer writing repo rules for Cline (VS Code 2026) and Windsurf (2026). Write rules from a stack brief. Do not invent package scripts. Not Cursor AGENTS.md, not CLAUDE.md, not Copilot instructions.md. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Tool I need (Cline, Windsurf, both): [Tool] - Repo name and languages: [Repo] - Package manager and versions: [Stack] - Commands that actually work: [Commands] - Directories the agent may edit: [Allow] - Directories the agent must not touch: [Deny] - Review bar before commit: [Review] - Words I must not use: [Banned] - Max lines: [LineCap or 200] Generate: 1. Tool lock for Cline and/or Windsurf 2026. Not Cursor, Claude Code, or Copilot. 2. Filename sketch: .clinerules and Windsurf rules path or UNKNOWN PATH. 3. Stack from Stack and Commands only. Missing scripts are NOT IN INPUTS. 4. Allow and Deny lists. Do not add paths. 5. Review bar from Commands plus Review. 6. Rules file draft under LineCap. 7. Compliance pass: quote Banned words and invented scripts. Cut them. Constraints: - Cline / Windsurf only. Do not emit AGENTS.md or CLAUDE.md. - Never invent a CLI command. - Stay under LineCap. - If Tool is Cline only, do not add Windsurf-only keys.
n8n Prompt Node Spec with Retry and Guardrails
Prompt node spec with retry and guardrails; version-lock n8n.
Act as an n8n automation engineer. Prompt node spec with retry and guardrails; version-lock n8n. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Tool and version: [Tool] - Job to automate: [Job] - Inputs/outputs: [IO] - Guardrails: [Guardrails] - Eval or retry needs: [Eval] - Forbidden: [Forbidden] Generate: 1. Job decomposition. 2. Tool-specific artifact (command, OpenAPI, Gem, scorer, n8n node). 3. Guardrails and retry. 4. Eval cases (5). 5. Gaps. Constraints: - No em dashes. - Never invent stats, citations, social proof, DOIs, or IDs. - If Inputs are thin, list gaps instead of filling. - Version-lock named tools; else write unknown. - De-identify and add not-advice for legal, clinical, insurance, HR, veterinary, or education-plan content. - Stay on brief for: n8n Prompt Node Spec with Retry and Guardrails
Tool-Call Transcript Scorer for an Agent Eval
Score tool-call transcripts; no invented gold tools.
Act as an agent-eval engineer. Score tool-call transcripts; no invented gold tools. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Tool and version: [Tool] - Job to automate: [Job] - Inputs/outputs: [IO] - Guardrails: [Guardrails] - Eval or retry needs: [Eval] - Forbidden: [Forbidden] Generate: 1. Job decomposition. 2. Tool-specific artifact (command, OpenAPI, Gem, scorer, n8n node). 3. Guardrails and retry. 4. Eval cases (5). 5. Gaps. Constraints: - No em dashes. - Never invent stats, citations, social proof, DOIs, or IDs. - If Inputs are thin, list gaps instead of filling. - Version-lock named tools; else write unknown. - De-identify and add not-advice for legal, clinical, insurance, HR, veterinary, or education-plan content. - Stay on brief for: Tool-Call Transcript Scorer for an Agent Eval
Port a Custom GPT Brief into a Gemini Gem
Port instructions; map never-list; do not invent Gemini-only features.
Act as an AI tools engineer porting GPT briefs to Gemini Gems. Port instructions; map never-list; do not invent Gemini-only features. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Tool and version: [Tool] - Job to automate: [Job] - Inputs/outputs: [IO] - Guardrails: [Guardrails] - Eval or retry needs: [Eval] - Forbidden: [Forbidden] Generate: 1. Job decomposition. 2. Tool-specific artifact (command, OpenAPI, Gem, scorer, n8n node). 3. Guardrails and retry. 4. Eval cases (5). 5. Gaps. Constraints: - No em dashes. - Never invent stats, citations, social proof, DOIs, or IDs. - If Inputs are thin, list gaps instead of filling. - Version-lock named tools; else write unknown. - De-identify and add not-advice for legal, clinical, insurance, HR, veterinary, or education-plan content. - Stay on brief for: Port a Custom GPT Brief into a Gemini Gem
GPT Actions OpenAPI Stub from a Job Description
OpenAPI stub for GPT Actions from a job description; no fake auth secrets.
Act as a Custom GPT Actions engineer. OpenAPI stub for GPT Actions from a job description; no fake auth secrets. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Tool and version: [Tool] - Job to automate: [Job] - Inputs/outputs: [IO] - Guardrails: [Guardrails] - Eval or retry needs: [Eval] - Forbidden: [Forbidden] Generate: 1. Job decomposition. 2. Tool-specific artifact (command, OpenAPI, Gem, scorer, n8n node). 3. Guardrails and retry. 4. Eval cases (5). 5. Gaps. Constraints: - No em dashes. - Never invent stats, citations, social proof, DOIs, or IDs. - If Inputs are thin, list gaps instead of filling. - Version-lock named tools; else write unknown. - De-identify and add not-advice for legal, clinical, insurance, HR, veterinary, or education-plan content. - Stay on brief for: GPT Actions OpenAPI Stub from a Job Description
Claude Code Slash Command from a Repeatable Job
Design a slash command for a repeatable job; no invented CLI flags.
Act as a Claude Code power user. Design a slash command for a repeatable job; no invented CLI flags. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Tool and version: [Tool] - Job to automate: [Job] - Inputs/outputs: [IO] - Guardrails: [Guardrails] - Eval or retry needs: [Eval] - Forbidden: [Forbidden] Generate: 1. Job decomposition. 2. Tool-specific artifact (command, OpenAPI, Gem, scorer, n8n node). 3. Guardrails and retry. 4. Eval cases (5). 5. Gaps. Constraints: - No em dashes. - Never invent stats, citations, social proof, DOIs, or IDs. - If Inputs are thin, list gaps instead of filling. - Version-lock named tools; else write unknown. - De-identify and add not-advice for legal, clinical, insurance, HR, veterinary, or education-plan content. - Stay on brief for: Claude Code Slash Command from a Repeatable Job
Tool-Call Agent Prompt-to-Eval Rubric
Turn an agent prompt into an eval rubric with tool-call gold traces, traps, and a pass/fail rule. No fake benchmark scores.
Act as an eval engineer. Turn an agent system prompt into a rubric plus cases. Distinct from the generic Prompt A/B harness and from RAG eval sets. Focus on tool calls: when to call, which args, when to refuse. Inputs: - Agent prompt: [Paste] - Tools (names + args): [Tools] - Failures we already saw: [Fails] - Must-not: [PII, payments, live prod] - Budget: [How many cases] Generate: 1. Behaviors to score: tool choice, args, refusal, final answer groundedness. 0-2 scale each. 2. Gold traces: for each case, the tool sequence (or none). 3. Cases: Budget count, mix of happy path, trap (should not call), missing-arg, tool error. 4. Decision rule: pass if mean meets a threshold you state. Fail the run on any must-not hit. Do not invent a public leaderboard score. 5. Trap ideas from Fails only plus obvious PII. 6. Refused: fabricating tool JSON the schema does not have. Constraints: - If Tools is empty, say this prompt cannot be tool-eval'd yet. - No fake 92 percent accuracy.
GitHub Copilot instructions.md Generator
Generate a GitHub Copilot instructions.md for a repo with stack, never-list, and test commands. Distinct from Cursor project rules.
Act as a staff engineer writing GitHub Copilot instructions.md. Distinct from Cursor project-rule files. Do not invent runtime versions. Inputs: - Repo purpose: [Repo] - Stack I can prove: [Stack] - Commands that work: [Commands] - Never-list: [Never] - Paths that need extra care: [Paths] - Secrets: [Secrets policy] Generate: 1. Map: Copilot instructions.md vs path notes. Do not emit Cursor mdc. 2. Full markdown: purpose, commands, style, never-list, secrets. 3. Path notes as headings, not YAML front matter. 4. Install: where the file goes. Humans fill versions. 5. Refused: print env files, disable safety, invent Node version. Constraints: - No Cursor globs YAML. - No fake CI badges. - Keep under 400 lines.
Support Agent System Prompt with Escalation and PII Rules
Write a support-agent system prompt with identity, scope, escalation, and PII/PHI rules, distinct from a reply playbook.
Act as a conversation-design lead. Write a system prompt for a support agent LLM. Distinct from the Customer Support Reply Playbook (that is a human reply pack). This is identity, tools, escalation, PII. Inputs: - Brand and channels: [Brand] - What the bot may do: [Scope] - Tools: [order lookup, none] - Escalation: [When, to whom] - PII/PHI rules: [What not to collect or echo] - Tone: [Tone] - Banned claims: [Banned] Generate: 1. System prompt (paste-ready): Identity, Scope, Style, Tools, Escalation, PII, Never-list, Closing. 2. Example turns (3): in-scope, escalate, PII refuse. 3. Tool-call sketch: when to call order lookup, what to do if tool fails. No fake APIs. 4. Distinct from playbook: do not write 10 canned emails. 5. Refused: asking for SSN, full card number, medical diagnosis. Constraints: - If Tools is none, the bot must not pretend to have looked up an order. - Do not leak other customers. - Keep the system prompt under 800 words.
Gemini Gem Instructions: Role, Goal, Input, Output
Write Gemini Gem instructions in Role / Goal / Input / Output / Constraints form with starters and a never-list.
Act as a Gemini Gem author. Output a Gem using Role, Goal, Input, Output, Constraints. Distinct from Custom GPT instructions and from Claude Project instructions. Inputs: - Gem name and job: [Job] - User: [User] - Gems tools they enabled: [Drive, browsing, none] - Never-list: [Never] - Examples they want inside the Gem: [Examples] Generate: 1. Gem name and one-line. 2. Instructions using exactly these headings: Role, Goal, Input, Output, Constraints (Never-list lives here). 3. Four example interactions (user / gem) that do not invent Drive file contents. 4. Tool policy: if browsing/Drive is off, refuse to pretend. 5. Port note back to Custom GPT (5 lines). 6. Refused: pretending to be Google support, exfiltrating Drive, medical orders. Constraints: - Headings must match Gemini's mental model: Role/Goal/Input/Output/Constraints. - Do not dump a 600-word Claude Project brief.
Custom GPT Instructions Builder: Role, Never-List, Starters
Build paste-ready Custom GPT instructions with a role, never-list, escalation, and conversation starters. Not a generic agent architect.
Act as a GPT Builder editor. Output paste-ready Custom GPT instructions. Distinct from the generic Custom AI Agent & System Prompt Architect and from Claude Project instructions. Inputs: - GPT name and job: [Job] - Who the user is: [User] - Knowledge files they will upload: [Files or none] - Never-list: [Never] - Escalation: [When to stop and tell them to use a human] - Tools: [browsing, code, image, none] - Starters I want: [Topics] Generate: 1. Name + one-line description (for the GPT store/profile, 200 chars or less). 2. Instructions block: Role, Audience, Process, Output format, Never-list, Escalation, Knowledge-file rules (cite file names, do not invent file contents). 3. Conversation starters (4), under 80 chars, matching Starters. 4. Tool notes: when to browse vs refuse if Tools is none. 5. Port note: 5 lines on what to change for a Gemini Gem vs a Custom GPT (do not output a full Gem unless asked). 6. Refused: jailbreak-welcome language, secret-exfiltration, medical prescriptions. Constraints: - Never-list must be in the instructions, not only in chat. - Do not invent knowledge-file quotes. - Distinct from a support-agent system prompt (that is another prompt in this batch).