Repository navigation
Releases: alibaba/skill-up
Releases · alibaba/skill-up
Release list
v0.14.0
Changelog
- 177ed34: ci: bump the github-actions group across 1 directory with 8 updates (#328) (@dependabot[bot])
- de025e6: feat(evaluator): run sequential judges in the shared case runtime (#315) (@zpzjzj)
- a337a8d: feat(judge): evaluate independent judges on one agent run (#290) (@zpzjzj)
- 87f86b5: feat(judge): support independent judge engine configuration (#314) (@zpzjzj)
- dd58a0f: fix(agent): use prompt files on Windows targets (#313) (@zpzjzj)
- 6d121cb: fix(ci): bound GoReleaser build concurrency on self-hosted runners (#323) (@zpzjzj)
- c9c41a0: fix(docs): use a local DeepWiki badge in READMEs (#322) (@zpzjzj)
- f84b842: fix(judge): normalize judge context host dir for relative --output-dir (#320) (@BiBoyang)
- 4d13bf1: fix(judge): prevent expect gate failures from inflating benchmark scores (#309) (@zpzjzj)
- 12deb53: fix(skill-upper): preserve intent and protect regression semantics (#291) (@zpzjzj)
Contributors
zpzjzj, BiBoyang, and dependabot
Assets 11
v0.13.0
Changelog
- 1063653: ci: bump node from
cd9f682to662933cin /action (#297) (@dependabot[bot]) - c4f4370: ci: bump the github-actions group with 3 updates (#299) (@dependabot[bot])
- 632cf5b: ci: bump the github-actions group with 4 updates (#268) (@dependabot[bot])
- df09a71: ci: bump the github-actions group with 6 updates (#285) (@dependabot[bot])
- 7f1ff9b: feat(agent): support OpenCode in local and sandbox runtimes (#275) (@zpzjzj)
- 35a8bef: feat(dsh): add approval-gated Skill observation loop (#270) (@zpzjzj)
- b85649b: feat(dsh): collect natural Skill follow-up feedback (#276) (@zpzjzj)
- c3cc4a8: feat(evaluator): simulate dynamic user turns (#286) (@zpzjzj)
- 1f489fb: feat(observation): add Codex Skill feedback plugin (#258) (@zpzjzj)
- 0c8bd9b: feat(plugin): add DeepSeek Harness skill evolution bundle (#254) (@zpzjzj)
- 3c5f722: feat(runtime): evaluate existing local workspace (#256) (@zpzjzj)
- 82fb23d: fix(action): extract pinned qodercli archive path (#248) (@zpzjzj)
- 889cfa6: fix(agent): handle cross-volume skill installs on Windows (#269) (@zpzjzj)
- d7bae74: fix(agent): resolve symlinked skill sources and fail loudly on zero-file installs (#257) (@BiBoyang)
- 42bc159: fix(agent): synthesize session-result output when a deadline kill leaves none (#265) (@BiBoyang)
- 5b0a2cd: fix(config): reject rule_based assertions with no recognizable matcher in validate (#295) (@BiBoyang)
- 6e7477a: fix(custom-engine): keep custom.kwargs out of the
SessionInputpayload (#289) (@Shxuuer) - 0579a86: fix(judge): preserve tool argument structure when matching (#304) (@mavericksea-ai)
- 4bf6d34: fix(qodercli): pass through explicit model names and IDs (#262) (@JHWang-1997)
- cd32b7d: fix(qodercli): pin 1.1.41 and raise self-eval concurrency (#278) (@zpzjzj)
Contributors
zpzjzj, BiBoyang, and 4 other contributors
Assets 11
v0.12.0
Changelog
- 3e7bd92: ci: bump actions/deploy-pages in the github-actions group (#245) (@dependabot[bot])
- b388301: ci: bump node from
367679ctocd9f682in /action (#244) (@dependabot[bot]) - 117c745: ci: require deterministic e2e suite on pull requests (#241) (@zpzjzj)
- 0186b47: feat(custom): support stateful multi-turn sessions (#250) (@zpzjzj)
- 3f2f82a: feat(examples): add DeepSeek Harness integration (#251) (@zpzjzj)
- 12ad944: fix(judge): harden first-attempt agent_judge prompt against prose prefixes (#260) (@JHWang-1997)
- 724d4fb: refactor(agent): retire legacy no-op config fields (#243) (@zpzjzj)
Contributors
zpzjzj, dependabot, and JHWang-1997
Assets 9
v0.11.0
Changelog
- cbffbb4: ci(qoder): route US e2e to regional endpoint (#227) (@zpzjzj)
- 36ed134: ci: bump the github-actions group with 3 updates (#235) (@dependabot[bot])
- 78acebc: feat(agent): add static runtime preflight (#233) (@zpzjzj)
- ebc7aa0: feat(cli): add evaluation event logs (#231) (@JHWang-1997)
- 8f3cf28: feat(cli): add explicit provider override (#222) (@zpzjzj)
- f10a8c9: feat(evalevent): add event protocol foundation (#229) (@JHWang-1997)
- e789437: fix(agent): isolate qodercli from inherited SDK mode (#242) (@JHWang-1997)
- e4e5df0: fix(e2e): restore Windows mock runner (#239) (@zpzjzj)
- 8051d18: refactor(agent): materialize protocol-aware model connections (#226) (@zpzjzj)
- 6eaac64: refactor(credential): add protocol-aware model connection primitives (#221) (@zpzjzj)
- 24b0077: refactor(evaluator): introduce execution plans (#228) (@JHWang-1997)
- acbbc74: test(e2e): ignore preflight in call counter (#237) (@zpzjzj)
Contributors
zpzjzj, dependabot, and JHWang-1997
Assets 9
v0.10.0
Changelog
- 2be51ef: ci: bump node from
cd56571to367679cin /action (#223) (@dependabot[bot]) - 35d439c: ci: bump the github-actions group with 3 updates (#224) (@dependabot[bot])
- 7e96b73: ci: bump the github-actions group with 5 updates (#218) (@dependabot[bot])
- cdc2751: feat(agent): support Qoder CLI CN edition (#216) (@zpzjzj)
- 4ea44b3: fix(qwen-code): exclude thought parts from transcript (#220) (@chiga0)
- 8566175: refactor(agent): centralize resolved configuration (#211) (@zpzjzj)
- a5dfe7f: refactor(agent): declare adapter capabilities (#217) (@zpzjzj)
Contributors
zpzjzj, chiga0, and dependabot
Assets 9
v0.9.1
Changelog
- 45c5a77: ci: bump node from
2d49d87tocd56571in /action (@dependabot[bot]) - 967a28d: ci: bump the github-actions group across 1 directory with 5 updates (@dependabot[bot])
- 70f4638: ci: use trusted Windows runner for manual E2E (@zpzjzj)
- a3c21b8: fix(agent): complete QoderCLI token validation (@JHWang-1997)
- 5d8d93e: fix(agent): handle large Codex JSONL records (@zpzjzj)
- c0e56cf: fix(agent): report QoderCLI token usage from JSON output (@JHWang-1997)
- 81f3d6e: fix(ci): support stock Windows PowerShell (@zpzjzj)
- 028c7f8: fix(cli): 更新 Judge debug 示例协议 (@DancingCircles)
- aed0cdd: fix(judge): isolate independent correction attempts (@DancingCircles)
- 422bdaf: fix(judge): preserve correction context across adapters (@DancingCircles)
- e055557: fix(judge): 修复围栏 JSON 字符串误判 (@DancingCircles)
- 7c9e1b1: fix(judge): 增加有界纠错重试 (@DancingCircles)
- 2b2d178: fix(judge): 强制严格 JSON 字段大小写 (@DancingCircles)
- 5face8b: fix(judge): 强化 agent_judge 严格输出协议 (@DancingCircles)
- b9d6bad: fix(judge): 拒绝重复 JSON 键并修正 Qoder E2E (@DancingCircles)
- 9a18eb6: fix(report): expose per-case execution metrics (@JHWang-1997)
- 97ca555: fix(report): retain failed judge metrics (@JHWang-1997)
- 7341109: fix(runner): scope report timing to each iteration (@JHWang-1997)
- 0e8752f: test(action): cover composed Codex configuration (@zpzjzj)
- 816f958: test(agent): characterize configuration compatibility (@zpzjzj)
Contributors
zpzjzj, dependabot, and 2 other contributors
Assets 9
v0.9.0
Changelog
- 25480cc: feat(judge): add output_matches rule (@googs1025)
- 71e9305: fix(agent): encode cwd markers as the CLIs write them (@roark47)
- 2d5127c: fix(agent): identify session transcripts by recorded cwd metadata (@roark47)
- 31b1ab0: fix(agent): locate session transcripts without CLI naming rules (@roark47)
- 3db0de2: fix(agent): never resume another workspace's session (@roark47)
- cdb13be: fix(ci): restore required pull request gate (@zpzjzj)
- 06d2dc4: fix(evaluator): scope each multi-turn response to its own turn (@roark47)
- 2153234: fix(judge): handle empty forbidden regex matches (@googs1025)
- 306e8cc: fix(release): install GitHub CLI before publish (@zpzjzj)
- 9a450c9: fix(runtime): drop summed length from env map capacity hint (@roark47)
- b1401f1: test(e2e): cover multi-turn session resume end to end (@roark47)
Contributors
zpzjzj, roark47, and googs1025
Assets 9
v0.8.0
Changelog
- 314f8c3: Initial plan (@Copilot)
- 0612f11: Potential fix for pull request finding (@roark47)
- 765b56b: Potential fix for pull request finding (@roark47)
- a4bc121: Potential fix for pull request finding (@roark47)
- 9595f5e: ci(e2e): lock Windows Claude CLI dependencies (@zpzjzj)
- 60c27a2: ci: address workflow review feedback (@roark47)
- 354b82d: ci: bump node from 22-bookworm-slim to 26-bookworm-slim in /action (@dependabot[bot])
- c74cc91: ci: bump the github-actions group across 1 directory with 4 updates (@dependabot[bot])
- 4f103ae: ci: bump the github-actions group with 2 updates (@dependabot[bot])
- faa8baf: ci: complete workflow maintenance hardening (@roark47)
- 86b2855: ci: harden GitHub Actions execution boundaries (@roark47)
- 9d5143f: ci: improve workflow release guarantees (@roark47)
- 6bb82f5: ci: pin workflow supply chain (@roark47)
- 5c4d44b: feat(cli): add baseline run flag (@googs1025)
- 5b2624c: feat(config): implement expect field in CaseDefaults with merge logic (@Copilot)
- 0a9404a: feat(config): support Skill file filters (@zpzjzj)
- 7fd4e0d: feat(report): add markdown report format (@googs1025)
- 984ea63: feat(run): summarize multi-iteration stability (@googs1025)
- 7668b48: fix(agent): convert Windows session paths before download (@zpzjzj)
- 89e0e95: fix(agent): match Claude Windows workspace sanitization (@zpzjzj)
- e3e640f: fix(agent): normalize Windows session workspace keys (@zpzjzj)
- 7f79f2e: fix(agent): preserve Windows workspace path spelling (@zpzjzj)
- f27b2c3: fix(ci): install go for codeql analysis (@roark47)
- 4b5355f: fix(ci): restore self-hosted runners (@roark47)
- ed94ec5: fix(ci): run required checks for docs-only changes (@zpzjzj)
- 83af16c: fix(ci): run zizmor without docker (@roark47)
- d6ddaf6: fix(config): satisfy perfsprint lint (@roark47)
- b0f3b9f: fix(evaluator): honor inherited exit code checks (@roark47)
- 821452a: fix(report): exclude without_skill baseline from aggregate stats (@JHWang-1997)
- f582843: fix(report): narrow markdown report scope (@googs1025)
- ada01a1: fix(report): prioritize explicit with-skill results (@JHWang-1997)
- 6cf5871: fix: address code review feedback - nil handling and redundant checks (@Copilot)
- 7cc5a3e: fix: address linting issues in test code (@Copilot)
- cfeb4e6: refactor(custom-engine): extract leaf config package (@zpzjzj)
- 4ce5038: refactor: pass default expect by value instead of pointer (@Copilot)
- ccb746a: test(config): complete default expect coverage (@roark47)
- dc4ae77: test(e2e): collect Windows Claude session diagnostics (@zpzjzj)
- 0065b4f: test(e2e): expose Windows session lookup diagnostics (@zpzjzj)
- 067f10f: test(e2e): verify Windows Claude token usage (@zpzjzj)
- d490116: test(evaluator): add integration tests for default expect behavior (@Copilot)
- bdae6c7: test(run): avoid unsafe ui output capture (@googs1025)
Contributors
zpzjzj, roark47, and 3 other contributors
Assets 9
v0.7.0
Changelog
- 3c013b1: Remove early evolution warning from README (@roark47)
- 916256b: Update README.zh.md to remove early stage warning (@roark47)
- a366f2f: feat(config): add per-case mocked MCP override schema and merge (@roark47)
- 939c20a: feat(evaluator): provision MCP config per case with override isolation (@roark47)
- 967125c: fix(config): validate per-case mcp overrides (@roark47)
- cb3a9d6: test(e2e): add per-case mocked MCP override eval fixture and test (@roark47)
Contributors
roark47
Assets 9
v0.6.0
Changelog
- fb41d22: Merge remote-tracking branch 'origin/main' into codex/implement-proposal-0002 (@roark47)
- dc801c2: Revert "docs: add judge agent skill usage guide" (@roark47)
- 9f8776c: ci: bump the github-actions group with 5 updates (@dependabot[bot])
- bdbd453: feat(agent): add SessionResumer interface for multi-turn conversation (@roark47)
- 8a37fd4: feat(agent): implement SessionResumer for Codex engine (@roark47)
- a93ce15: feat(config): add multi-turn conversation schema fields and validation (@roark47)
- 5037fb9: feat(evaluator): add multi-turn conversation execution engine (@roark47)
- 009979c: feat(evaluator): strengthen post_condition and capture semantics (@roark47)
- de841a9: feat(judge): add per-turn assertion rules for multi-turn evaluation (@roark47)
- a7a9196: feat(judge): support judge-specific skills (@roark47)
- 96a3422: feat(report): add per-turn results to report output formats (@roark47)
- 96a3c98: feat(report): surface judge skill metadata (@roark47)
- ab3dfa5: fix(agent): use codex exec resume for non-interactive session resumption (@roark47)
- caacdeb: fix(ci): adapt cleanup to runtime shell descriptor (@roark47)
- 93c6907: fix(evaluator): correct multi-turn execution accounting (@roark47)
- 1028b09: fix(evaluator): proceed to judge on skip_remaining post-condition failure (@roark47)
- 09f3684: fix(judge): address judge skill review feedback (@roark47)
- ca01982: fix(judge): include rule keywords in assertion evidence (@roark47)
- 488f0e9: fix(judge): require explicit judge skill invocation (@roark47)
- 5605b45: fix(report): render multi-turn conversation and failure reasons in HTML (@roark47)
- f972d95: fix: address copilot multi-turn review feedback (@roark47)
- dfaa69f: refactor(runtime): replace TargetGOOS with shell descriptor (@zpzjzj)
Contributors
zpzjzj, roark47, and dependabot
