agentlab-services

Several tools cluster around assessing output quality: attest_quality (signed QA pass/fail), honesty_audit (fabrication linter) and verify_text (spec-based quality check) have overlapping purposes, and screen_message vs endpoint_safety both do scam screening. The descriptions do differentiate targets (deliverable vs message vs endpoint), but an agent could still misselect among the quality-check trio.

All names are snake_case and readable, but the convention is mixed: verb_noun names (attest_quality, clean_csv, rank_opportunities, screen_message, verify_text) sit alongside noun-phrase/resource names (endpoint_safety, honesty_audit, market_brief, revenue_leak, trust_lease) and a bare verb (notarize). Predictable enough to scan, not a single clean pattern.

14 tools is squarely in the well-scoped range and each entry maps to a distinct service in the agent-economy workflow (attestation, opportunity/monetization, safety, utilities). No obvious filler or padding.

The attestation/notarization cluster (notarize, attest_quality, trust_lease, verify_text, honesty_audit) and monetization cluster (audit/rank/productize/revenue_*) are reasonably covered, but the set is heterogeneous and has gaps: no receipt verification/revocation, no listing of issued attestations, and clean_csv stands alone with no related data tools.