# v0.2.17 # { "Depends": "py-genlayer:1jb45aa8ynh2a9c9xn3b7qqh8sm5q93hwfp7jqmwsfhh8jpz09h6" } from genlayer import * from dataclasses import dataclass import json # --------------------------------------------------------------------- # Constants # --------------------------------------------------------------------- CATEGORY_UNVERIFIED = "unverified" CATEGORY_EMERGING = "emerging" CATEGORY_TRUSTED = "trusted" CATEGORY_EMINENT = "eminent" MAX_SCORE = 1000 MAX_CONFIDENCE = 1000 # represented in per-mille (0..1000) to stay integer MAX_HISTORY = 200 # per-profile rolling cap MAX_ENDORSEMENT_WEIGHT = 100 MAX_BIO_LEN = 280 MAX_DISPLAY_NAME_LEN = 80 MAX_NOTE_LEN = 200 MAX_REASON_LEN = 200 MAX_EXPLANATION_LEN = 800 MAX_SIGNALS_LEN = 4000 LLM_EQUIVALENCE_CRITERIA = ( "The two outputs are equivalent if both parse as JSON with the same keys " "(score, confidence, category, breakdown, explanation), the integer " "`score` values differ by at most 25, the `confidence` values differ by " "at most 100, and the `category` strings are identical." ) def _category_for_score(score: int) -> str: """Map a numeric 0-1000 score onto a trust category bucket.""" if score >= 800: return CATEGORY_EMINENT if score >= 500: return CATEGORY_TRUSTED if score >= 200: return CATEGORY_EMERGING return CATEGORY_UNVERIFIED def _clamp(value: int, lo: int, hi: int) -> int: if value < lo: return lo if value > hi: return hi return value # --------------------------------------------------------------------- # Storage records # --------------------------------------------------------------------- @allow_storage @dataclass class Profile: display_name: str bio: str score: u256 confidence: u256 # 0..1000 category: str last_evaluated_at: u256 created_at: u256 exists: bool @allow_storage @dataclass class ScoreEntry: score: u256 confidence: u256 category: str delta_abs: u256 delta_negative: bool reason: str explanation: str activity: u256 governance: u256 contribution: u256 trust: u256 created_at: u256 @allow_storage @dataclass class EndorsementRecord: from_addr: Address to_addr: Address weight: u256 note: str created_at: u256 revoked_at: u256 # 0 = active # --------------------------------------------------------------------- # Contract # --------------------------------------------------------------------- class Contract(gl.Contract): # --- single-value storage --- owner: Address version: u256 total_profiles: u256 total_evaluations: u256 total_endorsements: u256 # --- mappings --- profiles: TreeMap[Address, Profile] history: TreeMap[Address, DynArray[ScoreEntry]] eval_count: TreeMap[Address, u256] endorsements_given_keys: TreeMap[Address, DynArray[Address]] endorsements_received_keys: TreeMap[Address, DynArray[Address]] endorsements: TreeMap[Address, TreeMap[Address, EndorsementRecord]] # ================================================================= # Constructor # ================================================================= def __init__(self) -> None: self.owner = gl.message.sender_address self.version = u256(1) self.total_profiles = u256(0) self.total_evaluations = u256(0) self.total_endorsements = u256(0) # ================================================================= # Profile lifecycle (write) # ================================================================= @gl.public.write def register_profile(self, display_name: str, bio: str) -> None: """Create a profile for the caller. Idempotent: errors if already set.""" sender = gl.message.sender_address if sender in self.profiles: raise Exception("profile already exists for sender") display_name = display_name[:MAX_DISPLAY_NAME_LEN] bio = bio[:MAX_BIO_LEN] now = u256(gl.block.timestamp) self.profiles[sender] = Profile( display_name=display_name, bio=bio, score=u256(0), confidence=u256(0), category=CATEGORY_UNVERIFIED, last_evaluated_at=u256(0), created_at=now, exists=True, ) self.total_profiles = u256(self.total_profiles + 1) @gl.public.write def update_profile_metadata(self, display_name: str, bio: str) -> None: """Update the caller's display name / bio. Profile must exist.""" sender = gl.message.sender_address if sender not in self.profiles: raise Exception("profile not found") p = self.profiles[sender] p.display_name = display_name[:MAX_DISPLAY_NAME_LEN] p.bio = bio[:MAX_BIO_LEN] self.profiles[sender] = p # ================================================================= # AI evaluation (write — invokes Genlayer LLM under equivalence) # ================================================================= @gl.public.write def evaluate_and_update(self, target: Address, signals_json: str) -> None: """ Run the reputation engine over the supplied signal bundle and write a fresh score entry on-chain. `signals_json` is a JSON string the caller (typically the backend) assembled from wallet activity, governance, contributions, endorsements. The LLM call runs under the Genlayer equivalence principle so every validator independently re-derives a comparable score. """ if target not in self.profiles: raise Exception("profile not found") if len(signals_json) > MAX_SIGNALS_LEN: raise Exception("signals payload too large") prompt = ( "You are Reputon, an on-chain reputation engine. " "Given the following signal bundle for a wallet, output ONLY " "valid JSON (no prose, no markdown) with this exact schema:\n" '{\n' ' "score": ,\n' ' "confidence": ,\n' ' "category": <"unverified"|"emerging"|"trusted"|"eminent">,\n' ' "breakdown": {\n' ' "activity": ,\n' ' "governance": ,\n' ' "contribution": ,\n' ' "trust": \n' ' },\n' ' "explanation": \n' "}\n\n" "Scoring rules:\n" "- score must equal the sum of breakdown components, clamped to 0..1000.\n" "- category must match the score buckets: <200 unverified, " "200-499 emerging, 500-799 trusted, 800+ eminent.\n" "- be conservative when signals are sparse or contradictory; " "lower the confidence accordingly.\n" "- never invent activity that is not present in the signals.\n\n" f"Signal bundle:\n{signals_json}\n\n" "Respond with the JSON object only." ) def _ask_llm() -> str: return gl.nondet.exec_prompt(prompt) raw = gl.eq_principle.prompt_comparative(_ask_llm, LLM_EQUIVALENCE_CRITERIA) # Parse and validate try: decoded = json.loads(raw) except Exception: raise Exception("LLM produced non-JSON output") score = _clamp(int(decoded.get("score", 0)), 0, MAX_SCORE) confidence = _clamp(int(decoded.get("confidence", 0)), 0, MAX_CONFIDENCE) category = str(decoded.get("category", "")) or _category_for_score(score) if category not in ( CATEGORY_UNVERIFIED, CATEGORY_EMERGING, CATEGORY_TRUSTED, CATEGORY_EMINENT ): category = _category_for_score(score) bd = decoded.get("breakdown", {}) or {} activity = _clamp(int(bd.get("activity", 0)), 0, 250) governance = _clamp(int(bd.get("governance", 0)), 0, 250) contribution = _clamp(int(bd.get("contribution", 0)), 0, 250) trust = _clamp(int(bd.get("trust", 0)), 0, 250) explanation = str(decoded.get("explanation", ""))[:MAX_EXPLANATION_LEN] # Compute delta vs previous score prev = self.profiles[target] prev_score = int(prev.score) if score >= prev_score: delta_abs = u256(score - prev_score) delta_neg = False else: delta_abs = u256(prev_score - score) delta_neg = True now = u256(gl.block.timestamp) # Update profile snapshot prev.score = u256(score) prev.confidence = u256(confidence) prev.category = category prev.last_evaluated_at = now self.profiles[target] = prev # Append to history (rolling cap) if target not in self.history: self.history[target] = DynArray[ScoreEntry]() entries = self.history[target] entries.append(ScoreEntry( score=u256(score), confidence=u256(confidence), category=category, delta_abs=delta_abs, delta_negative=delta_neg, reason="evaluate_and_update", explanation=explanation, activity=u256(activity), governance=u256(governance), contribution=u256(contribution), trust=u256(trust), created_at=now, )) # Trim head if exceeded cap while len(entries) > MAX_HISTORY: entries.pop(0) self.history[target] = entries # Counters prev_count = u256(0) if target in self.eval_count: prev_count = self.eval_count[target] self.eval_count[target] = u256(prev_count + 1) self.total_evaluations = u256(self.total_evaluations + 1) # ================================================================= # Endorsements (write) # ================================================================= @gl.public.write def add_endorsement(self, target: Address, weight: int, note: str) -> None: """Caller endorses `target`. Repeated calls overwrite the prior entry.""" sender = gl.message.sender_address if sender == target: raise Exception("cannot endorse self") if sender not in self.profiles: raise Exception("endorser has no profile") if target not in self.profiles: raise Exception("target has no profile") w = _clamp(int(weight), 1, MAX_ENDORSEMENT_WEIGHT) note = note[:MAX_NOTE_LEN] now = u256(gl.block.timestamp) # Initialize per-sender map if sender not in self.endorsements: self.endorsements[sender] = TreeMap[Address, EndorsementRecord]() self.endorsements_given_keys[sender] = DynArray[Address]() if target not in self.endorsements_received_keys: self.endorsements_received_keys[target] = DynArray[Address]() sender_map = self.endorsements[sender] if target in sender_map: # update existing rec = sender_map[target] rec.weight = u256(w) rec.note = note rec.revoked_at = u256(0) sender_map[target] = rec else: sender_map[target] = EndorsementRecord( from_addr=sender, to_addr=target, weight=u256(w), note=note, created_at=now, revoked_at=u256(0), ) self.endorsements_given_keys[sender].append(target) self.endorsements_received_keys[target].append(sender) self.total_endorsements = u256(self.total_endorsements + 1) self.endorsements[sender] = sender_map @gl.public.write def revoke_endorsement(self, target: Address) -> None: """Caller revokes their existing endorsement for `target`.""" sender = gl.message.sender_address if sender not in self.endorsements: raise Exception("no endorsements from sender") sender_map = self.endorsements[sender] if target not in sender_map: raise Exception("no endorsement for target") rec = sender_map[target] if int(rec.revoked_at) != 0: raise Exception("already revoked") rec.revoked_at = u256(gl.block.timestamp) sender_map[target] = rec self.endorsements[sender] = sender_map # ================================================================= # Views # ================================================================= @gl.public.view def get_contract_info(self) -> dict: return { "name": "Reputon", "version": int(self.version), "owner": str(self.owner), "total_profiles": int(self.total_profiles), "total_evaluations": int(self.total_evaluations), "total_endorsements": int(self.total_endorsements), } @gl.public.view def has_profile(self, addr: Address) -> bool: return addr in self.profiles @gl.public.view def get_profile(self, addr: Address) -> dict: if addr not in self.profiles: raise Exception("profile not found") p = self.profiles[addr] return { "address": str(addr), "display_name": p.display_name, "bio": p.bio, "score": int(p.score), "confidence": int(p.confidence), "category": p.category, "last_evaluated_at": int(p.last_evaluated_at), "created_at": int(p.created_at), "evaluations": int(self.eval_count[addr]) if addr in self.eval_count else 0, } @gl.public.view def get_score(self, addr: Address) -> dict: if addr not in self.profiles: raise Exception("profile not found") p = self.profiles[addr] return { "address": str(addr), "score": int(p.score), "confidence": int(p.confidence), "category": p.category, "last_evaluated_at": int(p.last_evaluated_at), } @gl.public.view def verify_score(self, addr: Address, expected_score: int) -> bool: """Cheap exact-match verifier used by gating contracts off-chain.""" if addr not in self.profiles: return False return int(self.profiles[addr].score) == int(expected_score) @gl.public.view def get_history(self, addr: Address, limit: int) -> list: if addr not in self.history: return [] n = _clamp(int(limit), 1, MAX_HISTORY) entries = self.history[addr] out: list = [] count = len(entries) start = 0 if count <= n else count - n i = count - 1 while i >= start: e = entries[i] out.append({ "score": int(e.score), "confidence": int(e.confidence), "category": e.category, "delta": -int(e.delta_abs) if e.delta_negative else int(e.delta_abs), "reason": e.reason, "explanation": e.explanation, "breakdown": { "activity": int(e.activity), "governance": int(e.governance), "contribution": int(e.contribution), "trust": int(e.trust), }, "created_at": int(e.created_at), }) i -= 1 return out @gl.public.view def get_endorsements_given(self, addr: Address) -> list: out: list = [] if addr not in self.endorsements: return out sender_map = self.endorsements[addr] if addr not in self.endorsements_given_keys: return out keys = self.endorsements_given_keys[addr] i = 0 while i < len(keys): target = keys[i] if target in sender_map: rec = sender_map[target] out.append(_endorsement_to_dict(rec)) i += 1 return out @gl.public.view def get_endorsements_received(self, addr: Address) -> list: out: list = [] if addr not in self.endorsements_received_keys: return out senders = self.endorsements_received_keys[addr] i = 0 while i < len(senders): sender = senders[i] if sender in self.endorsements: sm = self.endorsements[sender] if addr in sm: out.append(_endorsement_to_dict(sm[addr])) i += 1 return out # --------------------------------------------------------------------- # Helpers (module-level — not part of contract storage) # --------------------------------------------------------------------- def _endorsement_to_dict(rec: EndorsementRecord) -> dict: return { "from": str(rec.from_addr), "to": str(rec.to_addr), "weight": int(rec.weight), "note": rec.note, "created_at": int(rec.created_at), "revoked_at": int(rec.revoked_at), "active": int(rec.revoked_at) == 0, }

Hash	Type	Status		From	To	Method	GenVM Result	Consensus Result	Time
0xd238c4b7...b4297a4a	Call	FINALIZED	IN	0x7401c1...1058Eb	0xD7975C...E3d911	register_profile	ERROR	Accepted	2 days ago
0xbaea6466...5bf17abb	Deploy	FINALIZED	IN	0x7401c1...1058Eb	0xD7975C...E3d911	(constructor)	SUCCESS	Accepted	3 days ago