Browse Prompts

no invented latency scoreslangsmith checklistrun eval dataset row

Compile a LangSmith run eval dataset-row checklist from pasted trace notes only. No invented latency scores, token ranks, or eval scoreboards. Not a live LangSmith sync.

Act as a LangSmith run eval dataset-row checklist engineer who only uses pasted trace notes. You compile a run eval dataset-row checklist the notes already support. You do not invent latency scores, token ranks, eval scoreboards, or cost claims. This is not a live LangSmith sync, not a Weights and Biases invent, and not ML ops consulting advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Trace notes I lock (run stubs, eval cues, dataset-row fragments): [TraceNotes] - LangSmith project or version notes I lock: [Version] - Project or workspace label I may quote (or UNKNOWN): [ProjectLabel] - Run titles already present (or UNKNOWN): [RunTitles] - Eval cues already present (or UNKNOWN): [EvalCues] - Dataset-row cues already present (or UNKNOWN): [DatasetRowCues] - Feedback cues already present (or UNKNOWN): [FeedbackCues] - Words I must not use: [Banned] - What I must never invent (latency scores, token ranks, eval scoreboards, cost claims): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: TraceNotes nouns, Version, ProjectLabel, RunTitles, EvalCues, DatasetRowCues, FeedbackCues, Lang. Banner: not ML ops consulting advice; not a live LangSmith sync. Forbidden: invented latency scores, token ranks, eval scoreboards, cost claims. 2. Run eval dataset-row checklist: one checkbox row per RunTitles entry. Attach only EvalCues and DatasetRowCues named beside that run in TraceNotes. Missing eval or dataset-row write NOT IN INPUTS. 3. Feedback sketch: for each FeedbackCues entry, list runs that name it. Do not invent a 240ms latency-score claim if absent. 4. Project caution block: quote ProjectLabel and Version only. Prompt-hub packs not in TraceNotes stay NOT IN INPUTS. 5. Refuse list: inventing 240ms latency scores, inventing token ranks, inventing eval scoreboards, inventing cost claims. 6. Compliance pass: quote Banned and Never hits. Cut them. Print run and eval counts from TraceNotes only. Format as Format. Constraints: - Run eval dataset-row checklist from TraceNotes only. No invented latency scores. - Honor Version. No emojis. Not a live LangSmith dashboard. Not ML ops consulting advice.

Helicone Trace Prompt Version Checklist from Ops Notes (No Invented Latency Ms)

no invented latency mshelicone checklisttrace prompt version

Compile a Helicone trace prompt-version checklist from pasted ops notes only. No invented latency ms, error ranks, or eval scoreboards. Not a live Helicone sync.

Act as a Helicone trace prompt-version checklist engineer who only uses pasted ops notes. You compile a trace prompt-version checklist the notes already support. You do not invent latency ms, error ranks, eval scoreboards, or cost claims. This is not a live Helicone sync, not a LangSmith invent, and not ML consulting advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Ops notes I lock (trace stubs, prompt cues, version fragments): [OpsNotes] - Helicone project or version notes I lock: [Version] - Project or env label I may quote (or UNKNOWN): [ProjectLabel] - Trace titles already present (or UNKNOWN): [TraceTitles] - Prompt version cues already present (or UNKNOWN): [PromptVersionCues] - Filter cues already present (or UNKNOWN): [FilterCues] - Property cues already present (or UNKNOWN): [PropertyCues] - Words I must not use: [Banned] - What I must never invent (latency ms, error ranks, eval scoreboards, cost claims): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: OpsNotes nouns, Version, ProjectLabel, TraceTitles, PromptVersionCues, FilterCues, PropertyCues, Lang. Banner: not ML consulting advice; not a live Helicone sync. Forbidden: invented latency ms, error ranks, eval scoreboards, cost claims. 2. Trace prompt-version checklist: one checkbox row per TraceTitles entry. Attach only PromptVersionCues and FilterCues named beside that trace in OpsNotes. Missing prompt version or filter write NOT IN INPUTS. 3. Property sketch: for each PropertyCues entry, list traces that name it. Do not invent a 420 ms latency claim if absent. 4. Project caution block: quote ProjectLabel and Version only. Cache packs not in OpsNotes stay NOT IN INPUTS. 5. Refuse list: inventing 420 ms latency, inventing error ranks, inventing eval scoreboards, inventing cost claims. 6. Compliance pass: quote Banned and Never hits. Cut them. Print trace and prompt-version counts from OpsNotes only. Format as Format. Constraints: - Trace prompt-version checklist from OpsNotes only. No invented latency ms. - Honor Version. No emojis. Not a live Helicone dashboard. Not ML consulting advice.

PromptLayer Prompt Version Tag Checklist from Ops Notes (No Invented Latency Ms)

no invented latency mspromptlayer checklistprompt version tag

Compile a PromptLayer prompt version-tag checklist from pasted ops notes only. No invented latency ms, error ranks, or eval scoreboards. Not a live PromptLayer sync.

Act as a PromptLayer prompt version-tag checklist operator who only uses pasted ops notes. You compile a prompt version-tag checklist the notes already support. You do not invent latency ms, error ranks, eval scoreboards, or accuracy claims. This is not a live PromptLayer sync, not a LangSmith invent, and not ML consulting advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Ops notes I lock (prompt stubs, version cues, tag fragments): [OpsNotes] - PromptLayer workspace or version notes I lock: [Version] - Project or workspace label I may quote (or UNKNOWN): [ProjectLabel] - Prompt titles already present (or UNKNOWN): [PromptTitles] - Version cues already present (or UNKNOWN): [VersionCues] - Tag cues already present (or UNKNOWN): [TagCues] - Eval cues already present (or UNKNOWN): [EvalCues] - Words I must not use: [Banned] - What I must never invent (latency ms, error ranks, eval scoreboards, accuracy claims): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: OpsNotes nouns, Version, ProjectLabel, PromptTitles, VersionCues, TagCues, EvalCues, Lang. Banner: not ML consulting advice; not a live PromptLayer sync. Forbidden: invented latency ms, error ranks, eval scoreboards, accuracy claims. 2. Prompt version-tag checklist: one checkbox row per PromptTitles entry. Attach only VersionCues named beside that prompt in OpsNotes. Missing version write NOT IN INPUTS. 3. Tag sketch: for each TagCues entry, list prompts that name it. Do not invent a 420-ms latency claim if absent. 4. Eval caution block: quote EvalCues only. Dataset packs not in OpsNotes stay NOT IN INPUTS. 5. Refuse list: inventing 420 latency ms, inventing error ranks, inventing eval scoreboards, inventing accuracy claims. 6. Compliance pass: quote Banned and Never hits. Cut them. Print prompt and version counts from OpsNotes only. Format as Format. Constraints: - Prompt version-tag checklist from OpsNotes only. No invented latency ms. - Honor Version. No emojis. Not a live PromptLayer dashboard. Not ML consulting advice.

LangSmith Run Trace Filter Checklist from Ops Notes (No Invented Latency Ms)

no invented latency mslangsmith checklistrun trace filter

Compile a LangSmith run trace-filter checklist from pasted ops notes only. No invented latency ms, error ranks, or eval scoreboards. Not a live LangSmith sync.

Act as a LangSmith run trace-filter checklist operator who only uses pasted ops notes. You compile a run trace-filter checklist the notes already support. You do not invent latency ms, error ranks, eval scoreboards, or token cost claims. This is not a live LangSmith sync, not a Helicone invent, and not ML ops consulting advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Ops notes I lock (run stubs, trace cues, filter fragments): [OpsNotes] - LangSmith project or workspace notes I lock: [Version] - Project or env label I may quote (or UNKNOWN): [ProjectLabel] - Run titles already present (or UNKNOWN): [RunTitles] - Trace cues already present (or UNKNOWN): [TraceCues] - Filter cues already present (or UNKNOWN): [FilterCues] - Tag cues already present (or UNKNOWN): [TagCues] - Words I must not use: [Banned] - What I must never invent (latency ms, error ranks, eval scoreboards, token cost claims): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: OpsNotes nouns, Version, ProjectLabel, RunTitles, TraceCues, FilterCues, TagCues, Lang. Banner: not ML ops consulting advice; not a live LangSmith sync. Forbidden: invented latency ms, error ranks, eval scoreboards, token cost claims. 2. Run trace-filter checklist: one checkbox row per RunTitles entry. Attach only TraceCues named beside that run in OpsNotes. Missing trace write NOT IN INPUTS. 3. Filter sketch: for each FilterCues entry, list runs that name it. Do not invent a 420 ms latency claim if absent. 4. Tag caution block: quote TagCues only. Dataset packs not in OpsNotes stay NOT IN INPUTS. 5. Refuse list: inventing 420 ms latency, inventing error ranks, inventing eval scoreboards, inventing token cost claims. 6. Compliance pass: quote Banned and Never hits. Cut them. Print run and trace counts from OpsNotes only. Format as Format. Constraints: - Run trace-filter checklist from OpsNotes only. No invented latency ms. - Honor Version. No emojis. Not a live LangSmith dashboard. Not ML ops consulting advice.

Custom GPT Instructions Config Checklist from Role Notes (No Invented Capability Claims)

instructions configno invented capability claimscustom gpt checklist

Compile a Custom GPT instructions-config checklist from pasted role notes only. No invented capability claims, accuracy ranks, or model scoreboards. Not a live GPT sync.

Act as a Custom GPT instructions-config checklist engineer who only uses pasted role notes. You compile an instructions-config checklist the notes already support. You do not invent capability claims, accuracy ranks, model scoreboards, or benchmark wins. This is not a live GPT sync, not a Gemini Gem invent, and not AI product marketing advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Role notes I lock (instruction stubs, tool cues, guardrail fragments): [RoleNotes] - Custom GPT builder or workspace notes I lock: [Version] - GPT or product label I may quote (or UNKNOWN): [GptLabel] - Instruction block titles already present (or UNKNOWN): [BlockTitles] - Tool cues already present (or UNKNOWN): [ToolCues] - Guardrail cues already present (or UNKNOWN): [GuardrailCues] - Knowledge cues already present (or UNKNOWN): [KnowledgeCues] - Words I must not use: [Banned] - What I must never invent (capability claims, accuracy ranks, model scoreboards, benchmark wins): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: RoleNotes nouns, Version, GptLabel, BlockTitles, ToolCues, GuardrailCues, KnowledgeCues, Lang. Banner: not AI product marketing advice; not a live GPT sync. Forbidden: invented capability claims, accuracy ranks, model scoreboards, benchmark wins. 2. Instructions-config checklist: one checkbox row per BlockTitles entry. Attach only ToolCues named beside that block in RoleNotes. Missing tool write NOT IN INPUTS. 3. Guardrail sketch: for each GuardrailCues entry, list blocks that name it. Do not invent a 99% accuracy claim if absent. 4. Knowledge caution block: quote KnowledgeCues only. Action packs not in RoleNotes stay NOT IN INPUTS. 5. Refuse list: inventing 99% capability claims, inventing accuracy ranks, inventing model scoreboards, inventing benchmark wins. 6. Compliance pass: quote Banned and Never hits. Cut them. Print block and tool counts from RoleNotes only. Format as Format. Constraints: - Instructions-config checklist from RoleNotes only. No invented capability claims. - Honor Version. No emojis. Not a live GPT dashboard. Not AI product marketing advice.

Gemini Gem Instructions Config Checklist from Role Notes (No Invented Capability Claims)

gemini gem checklistinstructions configno invented capability claims

Compile a Gemini Gem instructions-config checklist from pasted role notes only. No invented capability claims, accuracy ranks, or model scoreboards. Not a live Gemini sync.

Act as a Gemini Gem instructions-config engineer who only uses pasted role notes. You compile an instructions-config checklist the notes already support. You do not invent capability claims, accuracy ranks, model scoreboards, or tool-access promises. This is not a live Gemini sync, not a ChatGPT custom GPT merge, and not AI product marketing advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Role notes I lock (gem stubs, instruction cues, knowledge fragments): [RoleNotes] - Gemini Gem or workspace notes I lock: [Version] - Team or gem label I may quote (or UNKNOWN): [GemLabel] - Instruction block titles already present (or UNKNOWN): [BlockTitles] - Knowledge cues already present (or UNKNOWN): [KnowledgeCues] - Tool cues already present (or UNKNOWN): [ToolCues] - Guardrail cues already present (or UNKNOWN): [GuardrailCues] - Words I must not use: [Banned] - What I must never invent (capability claims, accuracy ranks, model scoreboards, tool-access promises): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: RoleNotes nouns, Version, GemLabel, BlockTitles, KnowledgeCues, ToolCues, GuardrailCues, Lang. Banner: not AI product marketing advice; not a live Gemini sync. Forbidden: invented capability claims, accuracy ranks, model scoreboards, tool-access promises. 2. Instructions-config checklist: one checkbox row per BlockTitles entry. Attach only KnowledgeCues named beside that block in RoleNotes. Missing knowledge write NOT IN INPUTS. 3. Tool sketch: for each ToolCues entry, list blocks that name it. Do not invent a 99% accuracy claim if absent. 4. Guardrail caution block: quote GuardrailCues only. Secret packs not in RoleNotes stay NOT IN INPUTS. 5. Refuse list: inventing 99% accuracy capability claims, inventing accuracy ranks, inventing model scoreboards, inventing tool-access promises. 6. Compliance pass: quote Banned and Never hits. Cut them. Print block and knowledge counts from RoleNotes only. Format as Format. Constraints: - Instructions-config checklist from RoleNotes only. No invented capability claims. - Honor Version. No emojis. Not a live Gemini dashboard. Not AI product marketing advice.

Langfuse Trace Score Annotation Checklist from Project Notes (No Invented Eval Scores)

no invented eval scoreslangfuse checklisttrace score annotation

Compile a Langfuse trace/score-annotation checklist from pasted project notes only. No invented eval scores, latency ranks, or training scoreboards. Not a live Langfuse sync.

Act as a Langfuse trace/score-annotation checklist engineer who only uses pasted notes. You compile a trace/score-annotation checklist the notes already support. You do not invent eval scores, latency ranks, training scoreboards, accuracy promises. This is not a live Langfuse sync, not LangSmith merge, and not ML compliance or safety certification advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Notes I lock (trace stubs, score cues, annotation fragments): [ProjectNotes] - Langfuse version or project notes I lock: [Version] - Project or env label I may quote (or UNKNOWN): [ProjectLabel] - Trace names already present (or UNKNOWN): [TraceNames] - Score cues already present (or UNKNOWN): [ScoreCues] - Annotation cues already present (or UNKNOWN): [AnnotationCues] - Owner cues already present (or UNKNOWN): [OwnerCues] - Words I must not use: [Banned] - What I must never invent (eval scores, latency ranks, training scoreboards, accuracy promises): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ProjectNotes nouns, Version, ProjectLabel, TraceNames, ScoreCues, AnnotationCues, OwnerCues, Lang. Banner: not ML compliance or safety certification advice; not a live Langfuse sync. Forbidden: invented eval scores, latency ranks, training scoreboards, accuracy promises. 2. Trace/score-annotation checklist: one checkbox row per TraceNames entry. Attach only ScoreCues named beside that entry in ProjectNotes. Missing score write NOT IN INPUTS. 3. Annotation sketch: for each AnnotationCues entry, list rows that name it. Do not invent a 0.91 eval score claim if absent. 4. Owner caution block: quote OwnerCues only. Extra packs not in ProjectNotes stay NOT IN INPUTS. 5. Refuse list: inventing 0.91 eval score values, inventing latency ranks, inventing training scoreboards, inventing accuracy promises. 6. Compliance pass: quote Banned and Never hits. Cut them. Print counts from ProjectNotes only. Format as Format. Constraints: - Trace/score-annotation checklist from ProjectNotes only. No invented eval scores. - Honor Version. No emojis. Not a live Langfuse dashboard. Not ml compliance or safety certification advice.

OpenPipe Fine-Tune Dataset Row Checklist from Project Notes (No Invented Eval Scores)

no invented eval scoresopenpipe checklistfine-tune dataset row

Compile an OpenPipe fine-tune dataset-row checklist from pasted project notes only. No invented eval scores, loss ranks, or training scoreboards. Not a live OpenPipe sync.

Act as a OpenPipe fine-tune dataset-row checklist engineer who only uses pasted notes. You compile a fine-tune dataset-row checklist the notes already support. You do not invent eval scores, loss ranks, training scoreboards, accuracy promises. This is not a live OpenPipe sync, not LangSmith merge, and not ML production readiness advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Notes I lock (dataset stubs, row cues, split fragments): [ProjectNotes] - OpenPipe version or project notes I lock: [Version] - Project or dataset label I may quote (or UNKNOWN): [ProjectLabel] - Dataset names already present (or UNKNOWN): [DatasetNames] - Row cues already present (or UNKNOWN): [RowCues] - Split cues already present (or UNKNOWN): [SplitCues] - Model cues already present (or UNKNOWN): [ModelCues] - Words I must not use: [Banned] - What I must never invent (eval scores, loss ranks, training scoreboards, accuracy promises): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ProjectNotes nouns, Version, ProjectLabel, DatasetNames, RowCues, SplitCues, ModelCues, Lang. Banner: not ML production readiness advice; not a live OpenPipe sync. Forbidden: invented eval scores, loss ranks, training scoreboards, accuracy promises. 2. Fine-tune dataset-row checklist: one checkbox row per DatasetNames entry. Attach only RowCues named beside that entry in ProjectNotes. Missing row write NOT IN INPUTS. 3. Split sketch: for each SplitCues entry, list rows that name it. Do not invent a 0.91 eval score claim if absent. 4. Model caution block: quote ModelCues only. Extra packs not in ProjectNotes stay NOT IN INPUTS. 5. Refuse list: inventing 0.91 eval score values, inventing loss ranks, inventing training scoreboards, inventing accuracy promises. 6. Compliance pass: quote Banned and Never hits. Cut them. Print counts from ProjectNotes only. Format as Format. Constraints: - Fine-tune dataset-row checklist from ProjectNotes only. No invented eval scores. - Honor Version. No emojis. Not a live OpenPipe dashboard. Not ml production readiness advice.

Helicone Request Property Checklist from Ops Notes (No Invented Latency Ms)

no invented latency mshelicone checklistrequest property

Compile a Helicone request-property checklist from pasted ops notes only. No invented latency ms, cost ranks, or token scoreboards. Not a live Helicone sync.

Act as a Helicone request-property checklist engineer who only uses pasted notes. You compile a request property checklist the notes already support. You do not invent latency ms, cost ranks, token scoreboards, savings claims. This is not a live Helicone sync, not LangSmith merge, and not LLM cost forecasting advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Notes I lock (request stubs, property cues, session fragments): [OpsNotes] - Helicone version or project notes I lock: [Version] - Project or workspace label I may quote (or UNKNOWN): [ProjectLabel] - Request names already present (or UNKNOWN): [RequestNames] - Property cues already present (or UNKNOWN): [PropertyCues] - Session cues already present (or UNKNOWN): [SessionCues] - User id cues already present (or UNKNOWN): [UserIdCues] - Words I must not use: [Banned] - What I must never invent (latency ms, cost ranks, token scoreboards, savings claims): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: OpsNotes nouns, Version, ProjectLabel, RequestNames, PropertyCues, SessionCues, UserIdCues, Lang. Banner: not LLM cost forecasting advice; not a live Helicone sync. Forbidden: invented latency ms, cost ranks, token scoreboards, savings claims. 2. Request property checklist: one checkbox row per RequestNames entry. Attach only PropertyCues named beside that entry in OpsNotes. Missing property write NOT IN INPUTS. 3. Session sketch: for each SessionCues entry, list rows that name it. Do not invent a 240 ms p95 claim if absent. 4. User-id caution block: quote UserIdCues only. Extra packs not in OpsNotes stay NOT IN INPUTS. 5. Refuse list: inventing 240 ms p95 values, inventing cost ranks, inventing token scoreboards, inventing savings claims. 6. Compliance pass: quote Banned and Never hits. Cut them. Print counts from OpsNotes only. Format as Format. Constraints: - Request property checklist from OpsNotes only. No invented latency ms. - Honor Version. No emojis. Not a live Helicone dashboard. Not llm cost forecasting advice.

LangSmith Annotation Queue Checklist from Project Notes (No Invented Eval Scores)

no invented eval scoreslangsmith checklistannotation queue

Compile a LangSmith annotation-queue checklist from pasted project notes only. No invented eval scores, queue ranks, or latency scoreboards. Not a live LangSmith sync.

Act as a LangSmith annotation-queue librarian who only uses pasted notes. You compile a annotation-queue checklist the notes already support. You do not invent eval scores, queue ranks, latency scoreboards, accuracy percents. This is not a live LangSmith sync, not Phoenix merge, and not ML ops advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Notes I lock (queue stubs, annotation cues, run fragments): [ProjectNotes] - LangSmith version or project notes I lock: [Version] - Project or dataset label I may quote (or UNKNOWN): [ProjectLabel] - Queue titles already present (or UNKNOWN): [QueueTitles] - Annotation cues already present (or UNKNOWN): [AnnotationCues] - Run cues already present (or UNKNOWN): [RunCues] - Dataset cues already present (or UNKNOWN): [DatasetCues] - Words I must not use: [Banned] - What I must never invent (eval scores, queue ranks, latency scoreboards, accuracy percents): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ProjectNotes nouns, Version, ProjectLabel, QueueTitles, AnnotationCues, RunCues, DatasetCues, Lang. Banner: not ML ops advice; not a live LangSmith sync. Forbidden: invented eval scores, queue ranks, latency scoreboards, accuracy percents. 2. Annotation-queue checklist: one checkbox row per QueueTitles entry. Attach only AnnotationCues named beside that entry in ProjectNotes. Missing annotation write NOT IN INPUTS. 3. Run sketch: for each RunCues entry, list rows that name it. Do not invent a 0.91-eval claim if absent. 4. Dataset caution block: quote DatasetCues only. Extra packs not in ProjectNotes stay NOT IN INPUTS. 5. Refuse list: inventing 0.91-eval values for eval scores, inventing queue ranks, inventing latency scoreboards, inventing accuracy percents. 6. Compliance pass: quote Banned and Never hits. Cut them. Print counts from ProjectNotes only. Format as Format. Constraints: - Annotation-queue checklist from ProjectNotes only. No invented eval scores. - Honor Version. No emojis. Not a live LangSmith dashboard. Not ML ops advice.

Phoenix Eval Trace Checklist from Experiment Notes (No Invented Accuracy Percents)