Remove the reputation lookup

[dmatlock96]

Summary

The AgentScore API no longer serves GET /v1/reputation, and the endpoint returns 404. This removes get_reputation, aget_reputation, the response types only they used, the README examples, and the reputation package keyword. The request path tests that used the lookup now use assess. The version goes to 3.0.0.

Type of change

• Bug fix (no breaking change)
• New feature (no breaking change)
• Breaking change (existing callers must update)
• Docs, tests, or internal maintenance only

Public API

Removed: AgentScore.get_reputation(address, chain=None), AgentScore.aget_reputation(address, chain=None), ReputationResponse, and the types that only it used: Grade, EntityType, ReputationStatus, Subject, Classification, Score, ChainScore, Identity, Activity, EvidenceSummary, Reputation, OperatorScore, AgentSummary, RedactedClassification, ChainEntry. VerificationLevel stays, as OperatorVerification uses it.

Migration: call assess(address, ...) or aassess(address, ...) and read decision and decision_reasons. There is no replacement for the score.

Test plan

• Deleted the lookup tests: the GET URL, the chain query, the fixture shape, and verification_level.
• Moved the request path tests (error status codes, invalid JSON, timeout, connect errors, auth header, User-Agent, context managers, close, concurrency, client reuse) to assess and aassess with a standalone ASSESS_PAYLOAD fixture.
• Deleted the live integration tests for the lookup.
• Ran uv run ruff check ., uv run ruff format --check ., uv run ty check agentscore/, uv run vulture . vulture_whitelist.py --min-confidence 80 --exclude .venv, and uv run pytest (174 passed, 2 skipped, 99.82% coverage). All pass.

Checklist

• Tests cover the new behavior, and the suite passes locally
• Lint, format, and type checks pass
• Docs and README examples updated if the public surface changed
• No secrets, credentials, or personal data in the diff or the tests

🤖 Generated with Claude Code