🤖 AI Tools
DSPy Signature Map from Module Inventory (No Invented Metric Scores)
Turn a DSPy module inventory into a signature map only. No invented metric scores, LM names, or dataset IDs beyond the inventory.
0Reviews
Prompt
Act as a DSPy program design engineer who only uses a pasted module inventory. You write a signature map the inventory already supports. You do not invent metric scores, LM names, dataset IDs, or optimizer traces. This is not a live DSPy compile run and not a vendor quote. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Module inventory I lock (signature stubs, field cues, predictor notes): [Inventory] - DSPy / Python package version notes I lock: [Version] - Lab or workspace label I may quote (or UNKNOWN): [Workspace] - Required signature or module names I may quote (or UNKNOWN): [SpecNames] - Signature class names already present (or UNKNOWN): [SignatureNames] - Input and output field names already present (or UNKNOWN): [FieldNames] - Metric names already present (or UNKNOWN): [MetricNames] - Words I must not use: [Banned] - What I must never invent (metric scores, LM names, dataset IDs, optimizer traces): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: Inventory nouns, Version, Workspace, SpecNames, SignatureNames, FieldNames, MetricNames, Lang. Forbidden: invented metric scores, LM names, dataset IDs, optimizer traces. Banner: not a live DSPy compile; not a vendor quote. 2. Signature map table: one row per Inventory signature stub or field cue. Missing predictor notes write NOT IN INPUTS. Use DSPy Signatures, Predictors, Metrics, and Teleprompters language when Inventory supports it. Quote SignatureNames and FieldNames only when present. 3. Spec name set: only names in SpecNames. Unnamed signatures stay NOT IN INPUTS. Never print metric-score or LM-name VALUES not in Inventory. 4. Version lock: print Version. Refuse DSPy features newer than Version if Version is named. 5. Refuse list: inventing metric scores, inventing LM names, inventing dataset IDs, inventing optimizer traces. 6. Compliance pass: quote Banned and Never hits. Cut them. Format as Format. Constraints: - Map from Inventory only. No invented metric-score VALUES. Teach DSPy signature mapping, not a generic LangChain swap. - Honor Version. No emojis.
Instructions
Replace every [bracket] with your details before running. Works on ChatGPT, Claude, and Gemini.
Generated Output
This image was generated using the prompt above.
Examples
Example Input
Inventory: signature Harbor Quay Answer Draft as pasted field cue question->answer as pasted predictor note Predict as pasted; signature River Ops Cite Check as pasted field cue claim->verdict as pasted. Metric score NOT IN INPUTS. LM name NONE. Version: DSPy 2.4 as pasted (do not invent unreleased MIPROv3 APIs) Workspace: Harbor Quay RAG Lab as pasted SpecNames: Answer Draft as pasted; Cite Check as pasted. Rerank Pass UNKNOWN. SignatureNames: AnswerDraft as pasted; CiteCheck as pasted. RerankPass UNKNOWN. FieldNames: question as pasted; answer as pasted; claim as pasted; verdict as pasted. context UNKNOWN. MetricNames: exact_match as pasted. f1 UNKNOWN. Banned: guaranteed score, LM invent, dataset invent, optimizer invent Never: invent metric scores, invent LM names, invent dataset IDs, invent optimizer traces Format: ledger + table + specs + version + refuse + compliance Lang: English
Example Output
1. Ledger. Inventory: signature Harbor Quay Answer Draft field cue question->answer predictor note Predict; signature River Ops Cite Check field cue claim->verdict. Metric score NOT IN INPUTS. LM name NONE. Version DSPy 2.4. Workspace Harbor Quay RAG Lab. SpecNames Answer Draft; Cite Check; Rerank Pass UNKNOWN. SignatureNames AnswerDraft; CiteCheck; RerankPass UNKNOWN. FieldNames question; answer; claim; verdict; context UNKNOWN. MetricNames exact_match; f1 UNKNOWN. Lang English. Banner: not a live DSPy compile; not a vendor quote. Forbidden: invented metric scores, LM names, dataset IDs, optimizer traces, guaranteed score, LM invent, dataset invent, optimizer invent. 2. Signature map. 1. Harbor Quay Answer Draft | signature AnswerDraft as pasted (SignatureNames) | fields question->answer as pasted (FieldNames) | predictor Predict as pasted | metric exact_match as pasted (MetricNames) | score NOT IN INPUTS | LM NONE | DSPy signature map only. 2. River Ops Cite Check | signature CiteCheck as pasted (SignatureNames) | fields claim->verdict as pasted (FieldNames) | predictor NOT IN INPUTS | metric NOT IN INPUTS | score NOT IN INPUTS | LM NONE. Dataset IDs not invented. Optimizer traces not printed. 3. Spec name set. Answer Draft; Cite Check as SpecNames. Rerank Pass UNKNOWN so write Rerank Pass NOT IN INPUTS. No metric-score VALUES printed. No third signature invented. 4. Version lock. DSPy 2.4 as pasted. Unreleased MIPROv3 APIs not used. Assertions NOT IN INPUTS. 5. Refuse. Score invent: refused. LM invent: refused. Dataset invent: refused. Optimizer invent: refused. Guaranteed score: refused. 6. Compliance. Banned hits none. Format ledger+table+specs+version+refuse+compliance. Gaps: predictor for River Ops Cite Check, Rerank Pass decision, context FieldNames, f1 MetricNames, RerankPass SignatureNames if any. Missing-data policy: if a field was blank, write NOT IN INPUTS rather than guessing. Lock any tool version named in Inputs; if unnamed, write unknown. No invented testimonials, star ratings, or press logos. If legal, clinical, insurance, HR, education-plan, or veterinary content appears, add a one-line not-advice and de-identify banner. Quote banned-word hits and cut them. End with a gaps list of five bullets the user still owes you. Character and byte caps in the job are hard; print counts when relevant. Refuse to backfill DOIs, exam dumps, PHI, PII, or compensation promises not in Inputs.