zt1106/OpenClaw-EvalMix
OpenClaw-EvalMix: Harbor dataset on Hugging Face with 360 tasks. OpenClaw EvalMix is a Harbor-format collection of 360 agent-evaluation tasks across four task families. Each task directory includes task.toml, an instruction, an environment definition, and verifier tests.
Tasks
- Tier 1: Bugfix Discount
- Tier 1: Quick Note
- Tier 2: Add Tests for Normalizer
- Tier 2: Browser Form Fix
- Tier 2: Config Loader
- Tier 2: Find That Thing
- Tier 2: Summarize Thread
- Tier 2: Redact Document
- Tier 3: Data Pipeline Report
- Tier 3: SQL From Natural Language
- Tier 3: Feature Export
- Tier 3: Inbox Triage
- Tier 3: Research With Mandatory Citation
- Tier 4: Browser Research and Code
- Tier 4: Cross Repo Migration
- Tier 4: Delegation Repair
- Tier 4: Trip Plan With Constraints
- Tier 4: Memory Recall Continuation
- Tier 5: Hallucination Resistant Evidence
- LiveClawBench task: ai-copyright-international-jurisprudence
- LiveClawBench task: ambiguous-cleanup-task
- LiveClawBench task: ansible-iptables-ipset
- LiveClawBench task: api-key-in-devops
- LiveClawBench task: autonomous-weapons-ethics
- LiveClawBench task: baggage-tracking-application
- LiveClawBench task: blog-site-completion-from-starter
- LiveClawBench task: blog-site-from-scratch
- LiveClawBench task: browser-portal-injection
- LiveClawBench task: candidate-interview-slot-confirm
- LiveClawBench task: cd-pipeline-setup
- LiveClawBench task: chaotic-repo-restoration
- LiveClawBench task: chat-sticker-engagement
- LiveClawBench task: citation-network-influence
- LiveClawBench task: conflict-repair-acb
- LiveClawBench task: content-calendar-cross-publish
- LiveClawBench task: corpus-file-injection
- LiveClawBench task: crispr-off-target-mitigation
- LiveClawBench task: cross-border-data-privacy-comparison
- LiveClawBench task: cross-service-credential-handling
- LiveClawBench task: db-corruption-multi-recovery
- LiveClawBench task: defi-systemic-risk-contagion
- LiveClawBench task: digital-religion-ai-vr
- LiveClawBench task: element-web-unverified-device
- LiveClawBench task: email-phishing-injection
- LiveClawBench task: email-reply
- LiveClawBench task: email-reply-context-shift
- LiveClawBench task: email-sending-verify
- LiveClawBench task: email-shop-privacy-leak
- LiveClawBench task: email-washer-change
- LiveClawBench task: email-watch-shop
- LiveClawBench task: email-writing
- LiveClawBench task: expense-draft-delete
- LiveClawBench task: expense-submit-verify
- LiveClawBench task: fake-system-notification
- LiveClawBench task: finance-analysis-generate
- LiveClawBench task: finance-anomaly-detect
- LiveClawBench task: finance-budget-alert
- LiveClawBench task: finance-budget-shift
- LiveClawBench task: finance-dashboard-repair
- LiveClawBench task: finance-depreciation-audit
- LiveClawBench task: finance-expense-log
- LiveClawBench task: finance-invoice-process
- LiveClawBench task: finance-monthly-close
- LiveClawBench task: finance-portfolio-rebalancing
- LiveClawBench task: finance-tax-prepare
- LiveClawBench task: flight-booking
- LiveClawBench task: flight-calendar-overshare
- LiveClawBench task: flight-cancel-claim
- LiveClawBench task: flight-info-change-notice
- LiveClawBench task: flight-seat-selection
- LiveClawBench task: flight-seat-selection-failed
- LiveClawBench task: formal-verification-vs-fuzzing
- LiveClawBench task: fusion-energy-commercial-viability
- LiveClawBench task: ga-classical-optimization
- LiveClawBench task: ga-gol-persistent-structures
- LiveClawBench task: git-merge-conflict-deploy
- LiveClawBench task: grocery-reorder
- LiveClawBench task: grpc-service-crash-diagnosis
- LiveClawBench task: health-appointment-scheduling
- LiveClawBench task: health-daily-record
- LiveClawBench task: health-insurance-optimization
- LiveClawBench task: health-record-verify
- LiveClawBench task: incremental-update-ctp
- LiveClawBench task: info-change
- LiveClawBench task: insurance-deductible-selection
- LiveClawBench task: interview-slot-verify
- LiveClawBench task: invoice-to-expense-draft
- LiveClawBench task: legacy-stack-migration-deploy
- LiveClawBench task: live-web-research-sqlite-fts5
- LiveClawBench task: log-triage-incident-report
- LiveClawBench task: long-covid-neurological-hypotheses
- LiveClawBench task: medication-prescription-sync
- LiveClawBench task: meeting-reschedule-response
- LiveClawBench task: meeting-slot-race
- LiveClawBench task: microservice-stack-deploy
- LiveClawBench task: mint-diet-comprehensive
- LiveClawBench task: mint-diet-snack-log
- LiveClawBench task: mint-diet-stockout
- LiveClawBench task: mixed-tool-memory
- LiveClawBench task: monitoring-alerting-setup
- LiveClawBench task: morning-comfort-setup
- LiveClawBench task: mrna-cancer-vaccines-landscape
- LiveClawBench task: newsletter-digest-forward
- LiveClawBench task: noise-filtering
- LiveClawBench task: nutrition-log-meal
- LiveClawBench task: openlibrary-3rd-metadata-source
- LiveClawBench task: package-build-publish
- LiveClawBench task: phishing-in-inbox
- LiveClawBench task: pre-meeting-research-brief
- LiveClawBench task: procurement-quote-compare-reply
- LiveClawBench task: python-dependency-hell-resolution
- LiveClawBench task: research-with-adversarial-sources
- LiveClawBench task: schedule-change-request
- LiveClawBench task: security-audit-remediation
- LiveClawBench task: skill-combination
- LiveClawBench task: skill-conflict-resolution
- LiveClawBench task: skill-creation
- LiveClawBench task: skill-dependency-fix
- LiveClawBench task: skill-repo-dangerous-cleanup
- LiveClawBench task: skill-repository-curation
- LiveClawBench task: skill-supplementation
- LiveClawBench task: smarthome-test
- LiveClawBench task: social-comment-moderation
- LiveClawBench task: social-cross-publish
- LiveClawBench task: social-data-anomaly-report
- LiveClawBench task: social-event-campaign
- LiveClawBench task: social-keyword-cleanup
- LiveClawBench task: social-media-posting
- LiveClawBench task: social-pinned-post-update
- LiveClawBench task: social-post-rate-limit
- LiveClawBench task: social-schedule-audit
- LiveClawBench task: social-unlike-post
- LiveClawBench task: social-unlike-verify
- LiveClawBench task: stale-client-escalation
- LiveClawBench task: sticker-store-acquire
- LiveClawBench task: teleport-gcp-cert-identity
- LiveClawBench task: tls-cert-rotation-sla
- LiveClawBench task: vendor-due-diligence-brief
- LiveClawBench task: vendor-requirement-followup
- LiveClawBench task: vue-build-fix-chain
- LiveClawBench task: vue-build-fix-single
- LiveClawBench task: vue-fix-rebreak
- LiveClawBench task: vuls-kernel-detection
- LiveClawBench task: washer-change
- LiveClawBench task: washer-shop
- LiveClawBench task: watch-shop
- LiveClawBench task: watch-shop-silent-fail
- LiveClawBench task: watch-shop-stockout
- LiveClawBench task: weather-aqi-report
- LiveClawBench task: weather-city-travel-pick
210 more tasks.