Changelog
What's new in OASM
Every release of the platform, straight from GitHub. New features, fixes and performance work, grouped so you can scan them fast.
New featuresBug fixesPerformanceRefactoringDocumentationOther changes
New features
- agentsNuclei-template-generator skill and save_template tool#853
- consoleAI gen button on Templates opens a prefilled AI Studio chat#854
- consoleTemplates page header, list filters and empty state#856
- consoleTemplate info bar, status bar and file menu in the editor#857
- vulnerabilitiesCheck AI evidence quotes against stored tool output#861
- agentsSubagent workflow orchestration and a Workflow tab#862
- consoleFolder explorer view for Files#863
- agentsRule-based tool guard and scope gate before approval#864
- consoleVertical agent workflow and Run AI Pentest#865
- Vulnerability insights on the dashboard#866
- adminAdmin IP allow-list enforced by an @Admin() guard#867
- workerWorker-local egress proxy for connector containers#850
- workerBuild version, WORKER_NAME and public IP/location on worker detail#871
Bug fixes
- agentsUse Responses API for OpenAI models#859
- workerResolve naabu targets through the worker's DNS#870
Refactoring
- consoleShare template upload and file actions#855
New features
- releaseStamp package.json files and commit before tagging
- vulnerabilitiesBulk approve and reject of AI proposals#848
- consoleBulk approve and reject of AI proposals in the vulnerability list#849
Documentation
- agentsAdd per-module AGENTS.md and document task release
New features
- agentsProject-scoped chats with project context, build tools and reasoning effort#834
- consoleUnify search with command palette and findings#835
- jobsShow run progress, ETA and retry#836
- templatesCustom Nuclei templates on workspace file storage#804
- vulnerabilitiesLifecycle management and generic issues#839
- consoleCluster layout for the asset graph with no node overlap#843
- targetsTarget overview, activity and settings tabs with a shared vulnerability view#844
Bug fixes
- agentsMatch the vulnerability and issue tools to the lifecycle model#841
- consolePolish the vulnerability list and detail pages#840
- assetsSort port, IP, technology and status code tabs by their column#842
- consoleAlign the vulnerability detail page and list spacing with the rest of the app#845
New features
- workersHint idle workers to claim jobs over the worker stream#814
- projectsRedesign projects with scope rules, triggers and provenance views#826
- workflowsTool inputs and outputs, and project triggers in the workflow file#827
- projectsSet schedules in the workflow editor, tab badges and a specific activity feed#828
- projectsRun project workflows on events and stop chains of runs#830
- projectsInstall tools from the workflow tool picker and audit tool installs#831
Bug fixes
- core-apiRecord the real client IP in the audit log behind a proxy#813
- workerReport a missing container runtime on connector jobs#825
- core-apiScan every discovered HTTP endpoint, not just the host#823
Refactoring
- projectsRename asset groups to projects and record vulnerability provenance#819
New features
- notificationsCursor pagination, infinite scroll and read toggles#781
- adminShow users' online status and last activity#789
- agentsAdd api-testing and osint-and-assets builtin skills#790
- agentsAdd specialist subagents for recon, web, network and validation#793
- agentsWiden subagent fan-out and give subagents short-term memory#794
- artifactsWorkspace file storage with sharing and quota#771
- agentsStop cancels subagents, resume mid-run, approval-gated plans#798
- agentsUpdate issue creation titles in bug and feature report workflows
- agentsHand subagent results to the next step and show every browser session#805
- consoleConfirm before opening external links#807
- core-apiRe-encrypt stored secrets with the active key at boot#810
- authSystem settings, gated sign-up with approval and email verification#811
Bug fixes
- remote-executeRemove unused stream endpoint lacking session authorization#780
- core-apiTake the workspace from the header, never from query params#784
- dashboardConsistent open counts, 0-100 security score and lighter queries#785
- core-apiGuard and confine the worker tool archive gRPC#808
- core-apiAuthenticated encryption that survives key rotation#809
Performance
- notificationsPatch cached pages instead of refetching after mutations#782
- authRedis session cache and consistent login, logout and workspace switching#783
- core-apiRun on the Node HTTP adapter instead of Express#786
Refactoring
- agentsStrengthen the builtin pentest skills#791
- agentsMake agent mode orchestrate instead of testing#792
Other changes
- Public sign-up, under your control. Admins can open sign-up from Admin → System → Authentication. New accounts confirm their email and wait for an admin to approve them. Sign-up is closed by default.
- Mail server settings. Configure SMTP under Admin → System → Mail server, with presets for Resend, SendGrid, Postmark, Mailgun, Amazon SES, Brevo and Gmail, and send a test email.
- User presence. Admins can see who is online and when each user was last active.
- Encryption. Stored credentials use authenticated AES-256-GCM tied to their key. Existing data is re-encrypted with your key at boot, and rotating keys is now safe.
- Required secrets. Deployments need
ENCRYPTION_KEYS and BETTER_AUTH_SECRET, so no stack runs on public default secrets anymore. - Tighter access.
- The workspace comes only from the
X-Workspace-Id header. - Worker file downloads are authenticated and confined to tool archives.
- An unused stream endpoint without session authorization was removed.
- External links. External links in the console ask before they open.
- Specialist subagents for reconnaissance, web, network and validation, with wider fan-out and short-term memory.
- Skills. New builtin skills for API testing and for OSINT and assets, and rewritten pentest skills that explain how, not just what.
- Run control.
- Stop cancels subagents too.
- A run can resume midway.
- Plans can require approval.
- Results and browser sessions. Subagent results reach the main agent without polling, and the Browser tab shows every browser session.
- Per-workspace file storage with sharing by email or public link, under a storage quota. Agents save their output files there.
- Faster API. Sessions are resolved from Redis, and the API runs on Node's HTTP adapter instead of Express.
- Notifications load page by page with infinite scroll and can be marked read or unread.
- Dashboard. Counts are consistent, the security score runs from 0 to 100, and the queries are lighter.
- Empty. This is every oasm-docker stack until now, whatever
.env said. Set a new key; data under the built-in default key stays readable. - A random value of 32+ characters. Keep it.
- A short or example value. Keep it first and add a new key after it:
ENCRYPTION_KEYS=<current key>,<new key>. - Everyone signs in again once, because of the new session secret.
- Public sign-up is closed by default. Admins can open it under Admin → System → Authentication once Admin → System → Mail server works. New accounts confirm their email and wait for an admin to approve them. Existing accounts, and accounts created by admins, are not affected.
- API clients send the workspace in the `X-Workspace-Id` header. A
workspaceId query parameter is no longer accepted, and DELETE /api/targets/:id/workspace/:workspaceId is now DELETE /api/targets/:id. - The first page load opens the first workspace, because no selection has been stored yet.
- Workers from v0.9.x keep working.
- Stored credentials were effectively unprotected on oasm-docker stacks. The compose file never passed
ENCRYPTION_KEYS, so credentials were encrypted with the built-in default key. After you set a key, they are re-encrypted with it (#809, #810). - Rotation could return wrong values. After a key rotation, about 0.4% of decrypts returned another value without an error. New data uses authenticated AES-256-GCM (#809).
- Sessions were signed with a public default. Without
BETTER_AUTH_SECRET, sessions and verification links were signed with a publicly known secret; it is now required (#811).
New features
- integrationsSend structured events to webhook integrations#745
- targetsAllow custom CIDR ranges from /24 to /32#748
- workerJoin core-api through Cloudflare over TLS#756
- agentsAdd OrcaRouter Cyber Inference as a first-class provider#750
- adminAdmin users and workspaces APIs and pages#767
- assetsDiscover URLs on demand and notify new URLs#772
- assetsTrack and badge assets found by the latest scan#773
- workerRun several workers on one host and scale the compose service#779
Bug fixes
- workerRun tool containers on hosts without CPU CFS quota#736
- workersRestrict global workers to admins#749
- vulnerabilitiesCascade dismissals when a vulnerability is deleted#755
- toolsMark installed tools as installed in installed list#759
- agentsStore each long-term memory as its own record#760
- adminSerialize first-admin init with a Redis lock#769
- depsPin patched transitive packages for security alerts#776
- workerBump github.com/docker/docker to v28.5.2#777
Refactoring
- adminMove admin logic into modules/admin under /admin/*#761
- core-apiAdd @Lock decorator and migrate distributed locks to it#770
- consoleKeep default table params out of the URL#774
New features
- targetsGit organization targets with repository secret scanning#688
- connectorAdd trivy scan repositories#694
- consoleAdd Create quick-action dropdown to header#695
- agentAllow agent to use connectors#690
- pr-sizeAdd workflow to enforce pull request size limits and guidelines
- toolsExpose agentTool flag and show agent icon in tools list#702
- consoleUnify app/settings sidebar and add AI Studio menu#704
- agentsPer-user agent config with models, web search and instruction settings#711
- vulnerabilitiesAdd JSON filter tree to vulnerability list#705
- consoleAdd filter builder to vulnerability list#706
- agentsAgentTool filter, landing tools tray and sidebar chat actions#712
- agentsTool install confirm dialog and landing prompt groups#713
- agentsAI Studio agents overhaul — custom agents, autonomous subagents, streaming resilience#716
- agentsResume runs on return, global workers for agents with sandboxed shell#719
- agentsResumable answer stream, split agent tools, AI Studio UI refresh#720
- toolsMulti-category tools with per-step category selection#715
- eventsSingle live events SSE stream with worker lifecycle events#723
- workersEvent-driven worker list with slim API and toolbar cleanup#724
- workersOne-time join keys, online-only agent workers#725
- vulnerabilitiesMulti-target fast scan, target logos and type badges#727
- consoleAdmin area, audit log filter toolbar and console UX polish#728
- consoleApp links in metadata and prefilled GitHub bug reports#729
- agentsRun agent to completion past the step limit#733
- integrationsNative Discord webhook notifications#732
- agentsSubagent transcripts with a chat side panel#739
- agentsResume runs on open, release streams on leave#740
- agentsAI Studio usage page and connectors page#741
- workerAgent-driven headless browser via "browser" commands#742
Bug fixes
- ciRegenerate pnpm-lock.yaml after dependency merges#701
- workerStore worker state under ~/.oasm instead of the working directory#703
- workerRestart pool wait budget whenever a slot is freed#731
- agentsClear stale approval prompts on resume#743
Refactoring
- consolecore-apiStructured asset filter builder#691
New features
- workflowsRun workflows as a needs-driven DAG#668
- jobs-registryWorkflow code view for a run#673
- agentsCommand approval system for agent tool execution#672
- core-apiEvent bus on Redis Streams#681
Bug fixes
- consoleGive the run DAG the worker graph canvas treatment#669
- core-apiClose cross-tenant, IDOR, SQLi and auth-bypass findings#671
Performance
- jobs-registryBatch job dispatch and fix quadratic fan-out#665
Refactoring
- consoleRevamp asset service detail page#683
- storagePresigned S3 uploads/downloads with per-bucket read authorization#680
- core-apiLayer the event bus and render integration messages from payloads#684
Documentation
- readmeRewrite README and add contributing and security policies
New features
- workerPush job cancellations over a bidirectional gRPC stream#664
Bug fixes
- ciRun console build natively to avoid QEMU SIGILL#661
- core-apiChunk vulnerability upsert under the Postgres parameter cap
- workflowsAlign vulnerability_scan_basic config with connector defaults
- core-apiKeep vulnerabilities when a workflow is deleted#662
- workerKeep the job poller alive and cap connector containers by resources#663
New features
- consoleMCP client selector with per-client configs#615
- workerNode Docker runtime with connector dispatch and pooling#621
- integrationAdd AWS integration features and improve core API structure#634
- toolsSupport urls discovery tool#639
- consoleAdd worker detail page with graph#640
- workerShow live running jobs in worker detail graph#644
- integrationAdd connect to Vercel and auto fetch domain, subdomain#641
- connectorRegister nikto, keep partial connector results, tidy vulnerability pages#649
- connectorAdd the ports_scanner category and richer job logging#650
- workerReconcile orphaned connector containers at startup#652
- workerAdd live worker runtime telemetry#653
- consoleGuide users to install tools for empty groups#654
Bug fixes
- consoleWarm workspace/current-permission at root for direct settings nav#616
- core-apiStore vulnerability severity in lowercase#642
- workerForward affected_url and description from findings#643
- consolePrevent Telegram buttons from submitting forms#647
- consoleAdd consistent skeleton loading states to dashboard cards#651
- core-apiPreserve asset group scan scope#656
- workerDetect docker.sock group gid for worker on Linux#655
- core-apiCreate missing bucket on upload 404#657
Refactoring
- workerInline gRPC client#617
- core-apiDrop connector tests, preserve masked secrets#635
Other changes
- @NienCongChua made their first contribution in https://github.com/oasm-platform/open-asm/pull/647