What's Changed

• feat(security): deployment-configurable extra command whitelist by @jacoblee-io in #316
• feat(cluster): consolidate cluster_info into cluster_list (search + structured metadata) by @jacoblee-io in #317
• fix: harden model fallback routing edge cases (#308 follow-up) by @LikiosSedo in #315

Full Changelog: v0.1.8...v0.1.9

Contributors

• @LikiosSedo
• @jacoblee-io

What's Changed

• feat(chat): scroll-to-select copy, image-safe copy & Markdown export by @a-yeyang in #300
• feat(tools+skills): background jobs, read-only HW diagnostics whitelist, skills cleanup by @jacoblee-io in #302
• feat(tools): background execution for host_exec + all script tools (+ sub-agent completion fixes) by @jacoblee-io in #303
• feat(tools): one-step pod-netns execution (node_exec/script + host_exec/script, kubectl & ssh) by @jacoblee-io in #304
• chore(skills): remove exec-smoke test fixture from skills/core by @jacoblee-io in #305
• fix(bg): don't wait for a paired server's completion — carve a generic server/client exception by @jacoblee-io in #306
• fix(stop): make the Stop button truly stop everything a session runs (bg jobs + fg exec + UI) by @jacoblee-io in #309
• feat: add pi-agent 0.78.1 model fallback routing by @LikiosSedo in #308
• feat(bg-jobs): status-aware background output (task_output tool + eager file + GC) by @jacoblee-io in #311
• Guard model fallback replay and compact preflight by @LikiosSedo in #312
• Remember chat agent and session selection by @LikiosSedo in #313
• Refine chat selection export controls by @LikiosSedo in #310

Full Changelog: v0.1.7...v0.1.8

Contributors

• @a-yeyang
• @LikiosSedo
• @jacoblee-io

What's Changed

• Add portal attachment OCR parsing by @LikiosSedo in #288
• fix(memory): disable Siclaw memory by default by @LikiosSedo in #289
• Sub-agents + Tasks-v2 plan ledger (background gated, fan-out concurrency capped) by @jacoblee-io in #290
• feat: managed (bastion-key) jump-host auth + host connection testing by @jacoblee-io in #294
• feat: multi-level SSH jump host (ProxyJump) support by @jacoblee-io in #293
• feat(skills): support skill directory packages by @LikiosSedo in #291
• fix(plan): allow parallel in_progress for fanned-out sub-agents; inline stays one-at-a-time by @jacoblee-io in #295
• fix(ui): auto-hide the plan panel ~5s after completion (live, not just on refresh) by @jacoblee-io in #296
• Keep Siclaw chat stable when agent selector expands by @LikiosSedo in #299
• feat(host): jump_chain login + host_list query (full host SSH capability) by @jacoblee-io in #297

Full Changelog: v0.1.6...v0.1.7

Contributors

• @LikiosSedo
• @jacoblee-io

What's Changed

• feat(portal-web): add light/dark theme toggle by @a-yeyang in #260
• feat(chat): per-message timing badges + Metrics latency breakdown by @a-yeyang in #259
• feat(chat): add copy buttons for user messages, tool commands, and tool outputs by @a-yeyang in #262
• Truncate chat session title fields before persistence by @LikiosSedo in #261
• Fix show time details some bugs by @a-yeyang in #263
• Support zip/tar skill pack upload with upsert semantics by @chent1996 in #264
• feat(runtime): make Runtime restart non-disruptive to active AgentBoxes by @chent1996 in #257
• Feat/create chart mcp by @a-yeyang in #265
• Improve chart renderer and fix message rendering bug by @a-yeyang in #266
• gateway->runtime by @a-yeyang in #268
• fix(portal): eliminate chart flicker during streaming by @a-yeyang in #269
• Feat/code block copy by @a-yeyang in #270
• Fix chart renderer label clipping and preview by @LikiosSedo in #271
• fix: derive developer-role support for compatible models by @LikiosSedo in #272
• fix: derive AgentBox gateway URL from runtime hostname by @LikiosSedo in #273
• Add Mermaid diagram rendering controls by @LikiosSedo in #274
• fix(test): eliminate flaky kubectl timeout in restricted-bash test by @a-yeyang in #277
• Fix empty streamed tool call arguments by @LikiosSedo in #275
• Feat/resizable import mcp dialog by @a-yeyang in #278
• Fix chat session title truncation by @LikiosSedo in #279
• feat(chat): add message edit-and-resend with input history navigation by @a-yeyang in #280
• Fix onboarding priority for operational tasks by @LikiosSedo in #281
• Fix/streaming cache stuck by @a-yeyang in #282
• Feat/mcp multi import by @a-yeyang in #283
• feat(agents): fork agent with hash-salted name by @a-yeyang in #285
• docs: align README audience tagline by @LikiosSedo in #287

Full Changelog: v0.1.5...v0.1.6

Contributors

• @a-yeyang
• @chent1996
• @LikiosSedo

What's Changed

• chore: bump version to 0.1.4 by @jacoblee-io in #152
• feat(frontend): add Dig deeper button for diagnostic depth control by @chent1996 in #153
• chore(scripts): add NFS session memory cleanup tool by @chent1996 in #156
• feat: Skill Space — collaborative skill sharing and management by @LikiosSedo in #154
• chore: remove outdated Chinese internal docs by @LikiosSedo in #160
• fix(security): prevent sensitive data leaking into AI context via kubectl by @chent1996 in #158
• fix(db): enable MySQL connection keep-alive by @jacoblee-io in #161
• Decouple Deep Investigation from normal mode with explicit state machine by @LikiosSedo in #155
• Fix skill fork path traversal error for relative skillsDir by @LikiosSedo in #163
• Fix skill diff: baseline tracking + personal fork eye icon by @LikiosSedo in #166
• fix(kb): persist document content in Gateway DB for vector recovery by @jacoblee-io in #164
• fix(tools): add kubectl rate protection for API server safety by @chent1996 in #165
• feat(security): add sensitive path protection for exec tools by @chent1996 in #162
• fix(skills): sync skill name between DB and SKILL.md frontmatter by @LikiosSedo in #170
• refactor(tools): restructure src/tools/ into 6 typed subdirectories by @chent1996 in #171
• refactor(prompt): streamline system prompt and add Understand Before Acting by @jacoblee-io in #169
• refactor(security): replace pre-execution blocking with post-execution sanitization by @chent1996 in #172
• docs(tools): add rate protection rules to tool descriptions by @chent1996 in #167
• fix(tools): align tool descriptions with actual runtime behavior by @chent1996 in #168
• refactor(tools): rename run_skill, stdin pipe, simplify netns tools by @jacoblee-io in #173
• fix(tools): handle backslash-escaped metacharacters in command parser by @jacoblee-io in #174
• docs(tools): sync docs with tool refactor, remove dead contexts by @jacoblee-io in #176
• fix(models): test connection reuses stored API key and verifies auth properly by @LikiosSedo in #181
• fix(web): reset hasLoadedRef on WS disconnect so pages reload on reconnect by @jacoblee-io in #178
• fix(db): provide explicit updatedAt in system_config insert by @jacoblee-io in #180
• fix(tools): support kubectl flag-before-subcommand and relax -A rate protection by @jacoblee-io in #183
• fix(tools): increase debug pod memory limit and remove requests by @jacoblee-io in #184
• refactor(skills): rename scope 'team' to 'global' across codebase by @LikiosSedo in #182
• fix(web): reconnect WebSocket when tab becomes visible by @jacoblee-io in #185
• refactor(security): unify command security model into COMMANDS + CONTEXT_POLICIES by @chent1996 in #189
• feat(memory): add workspace memory clearing by @chent1996 in #191
• refactor(tools): security pipeline facade + directory restructure by @chent1996 in #190
• feat(knowledge): batch upload and auto title extraction by @jacoblee-io in #192
• refactor(tools): directory cleanup + extract LLM utilities to shared by @chent1996 in #194
• refactor(skills): fork-edit-sync model with full lifecycle by @jacoblee-io in #188
• refactor(core): introduce guard pipeline to replace scattered monkey-patching by @chent1996 in #195
• docs: align harness framework with current codebase, add DDL parity test by @LikiosSedo in #193
• fix(cron): capture structured reports from models that produce empty text by @LikiosSedo in #198
• refactor(tools): introduce Tool Registry for declarative tool registration by @chent1996 in #196
• fix(security): remove POST and data flags from curl whitelist by @chent1996 in #200
• feat(web): make clear memory synchronous with success/error feedback by @chent1996 in #199
• fix(tools): remove kubectl exec from restricted-bash, consolidate to pod_exec/node_exec by @chent1996 in #197
• fix(core): add knowledgeDir to file read allowed directories by @jacoblee-io in #202
• fix(prompt): remove hardcoded 5-minute reference from cron section by @LikiosSedo in #201
• feat(models): add model editing support by @jacoblee-io in #203
• feat(cron): persist tool call traces for cron job executions by @LikiosSedo in #205
• fix(config): move debug pods to siclaw namespace by @chent1996 in #208
• fix(prompt): improve diagnostic completeness and stuck detection by @jacoblee-io in #206
• refactor(tools): replace skill management tools with skill_preview by @jacoblee-io in #207
• fix(web): fix steer pending message state management bugs by @jacoblee-io in #210
• fix(skills): forked personal/skillset skills now shadow their builtin origin by @jacoblee-io in #211
• fix(skills): check name conflict at contribute time and propagate approve errors by @jacoblee-io in #212
• fix(tools): add multi-cluster kubeconfig hint to bash tool description by @jacoblee-io in #213
• fix(web): replace hardcoded suggested prompts with rotating taglines by @LikiosSedo in #214
• fix(web): improve abort UX with visual feedback and prompt restore by @LikiosSedo in #215
• fix(cron): surface model errors in cron execution results and notifications by @jacoblee-io in #217
• docs: remove stale roadmap and outdated planning sections by @LikiosSedo in #219
• fix(tools): clarify that KUBECONFIG= env prefix is not supported in bash tool by @jacoblee-io in #220
• optimize(prompt): make knowledge/memory search on-demand by @jacoblee-io in #222
• feat(cron): add execution trace view to run history by @LikiosSedo in #221
• optimize(skill): add hardware diagnostic discovery to node-health-check by @jacoblee-io in #223
• fix(config): use default namespace for debug pods by @chent1996 in #224
• fix(brain): add empty response retry guard to pi-agent brain by @jacoblee-io in #225
• fix(prompt): enforce skill-first as hard gate before ad-hoc commands by @jacoblee-io in #227
• fix(command-sets): move -r/-R/--recursive block from category to per-command by @chent1996 in #226
• fix(tools): allow agent to read truncated output files in /tmp by @jacoblee-io in #228
• chore(oss): sync sanitized new commits from GitLab by @jacoblee-io in #235
• docs: align public docs with current architecture by @LikiosSedo in #236
• feat(portal-web): collapsible sidebar + promote Metrics to top-level by @LikiosSedo in #238
• feat(cli): unify TUI with local Portal via read-only snapshot by @LikiosSedo in #237
• docs(readme): add prominent Try-Demo link by @LikiosSedo in #240
• docs(readme): update architecture diagram to reflect current design by @LikiosSedo in #241
• refactor(helm): manage shared data PVC via chart by @chent1996 in #242
• refactor(runtime): drop legacy REST + jwtSecret; enforce singleton by @jacoblee-io in #245
• refactor(runtime): drop trusted-proxy + runtimeSecret + dead inbound /ws by @jacoblee-io in #248
• fix(internal-api): pass agentId to config.getMcpServers + getSkillBundle by @jacoblee-io in #249
• fix(local): wire knowledge sync, task notifications, and schedule card rendering by @LikiosSedo in #239
• refactor(dp): rebuild deep investigation on async delegation by @LikiosSedo in #247
• chore(helm...

Contributors

• @chent1996
• @LikiosSedo
• @jacoblee-io

What's Changed

• fix(gateway): SSO login button not visible after saving settings by @jacoblee-io in #112
• fix(web): fix metrics page chart sizing and route conflict by @chent1996 in #113
• feat: deep search quality gate, knowledge extraction, progressive discovery by @LikiosSedo in #111
• improve(skills): hint about test workspace when skill awaits approval by @jacoblee-io in #114
• feat: add workspace context directories (repos/ + docs/) with overview scanning by @LikiosSedo in #115
• fix: knowledge extraction schema + report paths by @LikiosSedo in #116
• fix(skills): Approvals page broken for reviewers viewing personal skills by @jacoblee-io in #117
• Fix non-admin users accessing admin-only pages and onboarding route leaks by @LikiosSedo in #119
• fix(db): force UTC timezone for mysql2 to fix audit query offset by @chent1996 in #120
• fix(db): force UTC timezone for mysql2 to fix audit query offset by @chent1996 in #121
• feat(cron): add rate limits and resource protection by @LikiosSedo in #122
• docs: align invariants/decisions/CLAUDE.md with implementation reality by @LikiosSedo in #125
• fix(ui): remove redundant environment selector from kubeconfig upload dialog by @jacoblee-io in #126
• fix(cron): use average interval and add tool-side pre-validation by @LikiosSedo in #124
• fix: allow combined short flags and add AUDIT_WRITE capability by @jacoblee-io in #127
• feat(tools): require explicit kubeconfig in multi-cluster sessions by @jacoblee-io in #128
• feat(memory): knowledge extraction pipeline improvements by @LikiosSedo in #129
• fix(deep-search): remove report file write that fails on K8s readOnlyRootFilesystem by @LikiosSedo in #131
• feat(deep-search): proportional budget allocation and enriched prior findings by @LikiosSedo in #130
• feat(agentbox): increase resource limits to 2c4g by @jacoblee-io in #133
• fix(web): clear cached user profile on logout by @jacoblee-io in #134
• refactor(dp): replace manage_checklist tool with system-event-driven checklist by @LikiosSedo in #136
• fix(web): fork_skill tool results not rendered in conversation UI by @jacoblee-io in #135
• fix(core): add compaction safeguard to prevent OOM during context compaction by @jacoblee-io in #137
• refactor(memory): simplify memory system — disable topic extraction and MEMORY.md injection by @LikiosSedo in #141
• fix(core): add tool call validation to prevent 400 errors from OpenAI-compatible APIs by @jacoblee-io in #140
• fix(deep-search): catch Phase 1 sub-agent timeout to prevent investigation failure by @LikiosSedo in #139
• fix(tui): clear hypothesis widget before opening select dialog by @LikiosSedo in #138
• feat(web,db,tools): add cluster infra context and rename environments to clusters by @jacoblee-io in #144
• feat(feedback): add interactive session feedback system by @LikiosSedo in #142
• feat: harden debug pod lifecycle — labels, cleanup, namespace, reuse, GC by @chent1996 in #145
• feat(knowledge): add team-level knowledge base by @jacoblee-io in #146
• fix(skills): auto-extract description from frontmatter, fix script deletion by @jacoblee-io in #147
• fix(web): prevent DP mode from persisting across login sessions by @LikiosSedo in #148
• feat(debug-pod): per-cluster debug image and cache isolation by @chent1996 in #149
• fix(db): add missing feedback_reports table to MySQL DDL by @LikiosSedo in #150

Full Changelog: v0.1.3...v0.1.4

Contributors

• @chent1996
• @LikiosSedo
• @jacoblee-io

What's Changed

• refactor(k8s): replace per-user dynamic PVCs with shared PVC + subPath by @jacoblee-io in #105
• fix(k8s): remove stale NFS writes and decouple PVC from skills by @jacoblee-io in #106
• fix(gateway): remove non-root user and fix debugImage propagation by @jacoblee-io in #108
• refactor(helm): clean up pseudo-configurable fields and unify image config by @jacoblee-io in #109
• chore: bump version to 0.1.3 by @jacoblee-io in #110

Full Changelog: v0.1.2...v0.1.3

Contributors

• @jacoblee-io

What's Changed

• chore: sync package-lock.json version to 0.1.1 by @nightmeng in #43
• docs: add demo GIF to README by @LikiosSedo in #44
• fix: survive WS disconnect during deep investigation by @LikiosSedo in #45
• fix: ensure positive X.509 serial number in mTLS certificates by @nightmeng in #46
• docs: update install commands, add demo GIF, reduce redundancy by @LikiosSedo in #47
• fix: TUI first-run bugs and onboarding redesign by @LikiosSedo in #48
• docs: update README with Slack link and fix install commands by @LikiosSedo in #50
• fix: TUI first-run bugs and kubeconfig management by @LikiosSedo in #49
• fix: TUI deep investigation mode activation and interactive hypotheses by @LikiosSedo in #51
• fix: replace obsolete mcp-servers.json fallback with settings.json by @nightmeng in #53
• Dev/ob by @chent1996 in #52
• chore(ui): hide Brains menu until claude-sdk is ready by @nightmeng in #55
• refactor: remove env var config delivery for AgentBox by @nightmeng in #57
• feat: add /setup in-session command and first-run wizard for TUI mode by @LikiosSedo in #56
• feat: built-in metrics dashboard with dual-mode support by @chent1996 in #54
• Align public docs with current product behavior by @LikiosSedo in #58
• fix: TUI session switching bugs (loading state, DP restore, scroll, race protection) by @LikiosSedo in #59
• fix(cron): pass workspace envId to schedule card and panel by @chent1996 in #61
• fix: skill AI review — URL, spec coverage, config caching, and UX by @LikiosSedo in #63
• feat(metrics): add skill call metrics across full stack by @chent1996 in #60
• security: defense-in-depth for AgentBox command execution by @nightmeng in #64
• refactor(settings): clean up System Settings — remove unused configs by @nightmeng in #65
• fix(ui): tooltip clipped on skill review cards by @nightmeng in #66
• fix(prompt): use mode-aware setup guidance for Web UI vs TUI by @LikiosSedo in #68
• fix(agentbox): resolve MCP tools not loading on readOnlyRootFilesystem by @chent1996 in #67
• feat: unified Credentials page with Kubernetes/SSH/API tabs by @nightmeng in #69
• feat(audit): command audit system for tool execution tracking by @chent1996 in #71
• feat(k8s): remove init container + add PVC persistence for user memory by @nightmeng in #72
• feat(cron): robustness improvements — retry, locks, workspace, indexes by @nightmeng in #70
• fix: improve new user session UX — language, credentials, profile by @LikiosSedo in #75
• feat(db): add scheduled cleanup for sessions, messages, and session_stats by @chent1996 in #73
• fix: improve skill tool descriptions and remove Skill Scripts Reference by @nightmeng in #76
• fix(ui): workspace switch & notification z-index by @nightmeng in #78
• fix: restrict schedule and skill tools to web/channel mode by @LikiosSedo in #80
• fix: deep investigation UI — IME conflict, refresh state restoration by @LikiosSedo in #79
• feat(skills): add quota-debug and volcano-queue-debug diagnostic skills by @chent1996 in #77
• feat(deep-search): 3-layer prior knowledge injection for hypothesis generation & validation by @LikiosSedo in #74
• feat(skills): add Volcano scheduling diagnostic skills for SRE by @wangao1236 in #83
• fix: embedding config save button and API request by @nightmeng in #84
• fix: make skills/extension/ discoverable by agent and Gateway by @nightmeng in #85
• fix: script-resolver builtin fallback should search skills/extension/ by @nightmeng in #87
• fix: deep_search Phase 3 sub-agents silently produce 0 tool calls in TUI mode by @LikiosSedo in #91
• fix: correct hypothesis parsing in TUI widget for indented and multi-line formats by @LikiosSedo in #89
• feat(skills): add networkpolicy-debug diagnostic skill by @chent1996 in #90
• fix: add skill draft mode for TUI, prevent silent skill creation by @LikiosSedo in #88
• docs: add skills system design document by @nightmeng in #86
• fix: improve Web UI empty state and investigation exit UX by @LikiosSedo in #92
• fix: address 19 bugs from deep code review by @nightmeng in #93
• chore: remove unused @aws-sdk/client-s3 dependency by @jacoblee-io in #97
• fix: structured output for propose_hypotheses to fix hypothesis card title bleed-through by @LikiosSedo in #94
• feat: copy extension skills into Docker images by @jacoblee-io in #99
• fix: correct local-mode skill sync docs and add embedding batch limit by @LikiosSedo in #96
• fix(web): auto-search audit logs when filter selection changes by @chent1996 in #95
• feat: support custom debugImage in Helm chart by @jacoblee-io in #102
• feat: investigation feedback mechanism + fix DP checklist auto-complete by @LikiosSedo in #101
• fix(security): eliminate command injection in get-node-logs.sh by @chent1996 in #100
• refactor(gateway): merge standalone cron process into gateway by @jacoblee-io in #103
• feat: add TUI feedback prompt after deep investigation by @LikiosSedo in #104

New Contributors

• @wangao1236 made their first contribution in #83
• @jacoblee-io made their first contribution in #97

Full Changelog: v0.1.1...v0.1.2

Contributors

• @nightmeng
• @wangao1236
• @chent1996
• @LikiosSedo
• @jacoblee-io

What's Changed

• docs: expand roadmap with OB, CS tracks and trim speculative phases by @LikiosSedo in #36
• feat(gateway): redesign Model API configuration with Quick Setup and provider presets by @LikiosSedo in #37
• chore: update default docker registry to siclaw by @nightmeng in #40
• fix: resolve built-in skills from package root instead of cwd by @LikiosSedo in #41
• chore: bump version to 0.1.1 by @nightmeng in #42

Full Changelog: v0.1.0...v0.1.1

Contributors

• @nightmeng
• @LikiosSedo

What's Changed

• Initial commit for siclaw by @fred-scitix in #1
• fix: restore MCP mounts and config loading for agentbox pods by @chent1996 in #2
• feat: improve deep search engine and harden kubeconfig propagation by @LikiosSedo in #4
• refactor: decouple deep investigation capability from UI toggle state by @LikiosSedo in #6
• feat: add explicit [+ Add] button to KVEditor for multiple key-value ... by @chent1996 in #5
• feat: skills system optimization - DB storage, mTLS, S3 removal by @nightmeng in #9
• feat(tools,gateway): credential output protection and DP decouple by @LikiosSedo in #10
• feat: multi-tab isolation, stateless Gateway, AgentBox self-governance by @nightmeng in #11
• fix: harden command execution security with whitelist-only validation by @nightmeng in #14
• feat(deep-search): investigation memory loop for hypothesis enhancement by @LikiosSedo in #15
• feat: restrict file tools + adaptive investigation by @LikiosSedo in #13
• refactor: remove NFS dependency for MCP config sync by @chent1996 in #16
• feat: structured knowledge extraction from deep investigation conclusions by @LikiosSedo in #17
• feat: skill UX improvements, label editing, and tool validation by @nightmeng in #18
• refactor: unify Gateway-AgentBox resource sync for MCP and Skills by @chent1996 in #20
• fix: align tests with LOCAL_BLOCKED_COMMANDS security hardening by @nightmeng in #27
• feat: add unified siclaw CLI entry for npx support by @nightmeng in #28
• chore: add community infrastructure (CI, templates, CoC, security) by @nightmeng in #21
• refactor(ui): reorganize sidebar menu layout by @nightmeng in #30
• docs: add Claude operating manual and architecture design docs by @LikiosSedo in #31
• refactor: declarative command rule engine + remove legacy MCP config by @nightmeng in #32
• Fix/local spawner resource sync by @chent1996 in #29
• docs: add 5-layer system architecture diagram to README by @LikiosSedo in #35
• docs: add Mintlify documentation site by @LikiosSedo in #38
• chore: prepare for npm publish by @nightmeng in #39

New Contributors

• @fred-scitix made their first contribution in #1
• @chent1996 made their first contribution in #2
• @LikiosSedo made their first contribution in #4
• @nightmeng made their first contribution in #9

Full Changelog: https://github.com/scitix/siclaw/commits/v0.1.0

Contributors

• @nightmeng
• @chent1996
• @LikiosSedo
• @fred-scitix