Changelog

Product updates, reliability work, interface improvements, and release notes for the MagicSword.

Showing 1-10 of 77 releases

November 12, 2026Since v1.1.68

v1.1.69 includes 1 update since v1.1.68. The portal now publishes the November 5 and November 12 Living Off the Land (LOTL) weekly editions, restores the missed October 29–November 5 edition as retrospectively verified, and adds 18 new credits and two lifetime proofs from an expanded 37-project inventory.

Highlights

  • Browse the new November 5–12 LOTL weekly edition and the restored October 29–November 5 edition.
  • Expanded LOTL inventory to 37 projects, including 34 public repositories across 32 audited projects.
  • Added 18 new credits and two lifetime proofs; annual rankings now show 960 credits from 176 people.
  • All 951 prior credit records, 200 proofs, and 37 project source mappings preserved exactly.
  • All 11 LOTL editions validated with 142 tests for accurate, consistent weekly content.

Features

  • Browse the November 5–12 LOTL weekly edition: You can now read the latest Living Off the Land (LOTL) weekly edition, covering 16 PRs across LOOBins and LOLESXi from the November 5–12 interval and including a verified first-time contributor.
November 11, 2026Since v1.1.67

v1.1.68 includes 5 updates since v1.1.67.

Highlights

  • Add MagicSword marketing brand guide
  • Preserve policy decisions during intelligence source updates
  • Improve exception review clarity, safe saving, and Event Review access
  • Keep ExtSentry browsable on Free accounts

Improvements

  • Add MagicSword marketing brand guide
  • Improve exception review clarity, safe saving, and Event Review access
  • Keep ExtSentry browsable on Free accounts

Fixes

  • Preserve policy decisions during intelligence source updates
  • Fix sword replay coverage race for release 1.1.68
November 10, 2026Since v1.1.66

v1.1.67 includes 6 updates since v1.1.66. This release fixes API key creation for organization-funded SSO admins, adds publisher-based MSI approvals with consistent signer metadata, and surfaces matching deny rules before you save a policy exception.

Highlights

  • Organization-funded SSO admins can now create API keys without a separate personal grant, with inline errors and retry.
  • Signed MSI events can be approved by publisher, with signer metadata shown consistently across Investigate, Event Review, CSV, and AI summaries.
  • Policy exceptions now surface matching deny rules before saving, so you can resolve the actual block instead of creating an ineffective Allow.
  • New actively exploited code signing certificates have been added to threat intelligence.
  • The threat intelligence page now leads with MagicSword community research and tools.

Fixes

  • Organization-funded SSO admins can create API keys: You can now create API keys as an SSO admin in an organization-funded account without needing a separate personal grant. Key creation and list errors appear inline with retry, and revocation feedback is reliable.
  • Review matching deny rules before applying policy exceptions: You can now see the deny rules that still block a file before saving an Allow, so you don't create an exception that leaves the block in place. Only explicitly selected customer-owned denies are changed, and broad changes require acknowledgement.
  • Review matching deny rules before applying policy exceptions: You can now see the deny rules that still block a file before saving an Allow, so you don't create an exception that leaves the block in place. Only explicitly selected customer-owned denies are changed, and broad changes require acknowledgement.

Features

  • Approve signed MSI events by publisher: You can now approve a signed MSI by its publisher instead of only by file hash, so updates from the same trusted publisher stay allowed without new hashes. Unsigned or ambiguous MSI events still default to Hash.

Security

  • New actively exploited signer intelligence: New actively exploited code signing certificates have been added to threat intelligence, so you can identify and block signers known to be abused in active attacks.

Improvements

  • Research projects moved higher on threat intelligence page: The threat intelligence page now leads with MagicSword community research and tools directly below the hero, so you can find the latest research without scrolling.
November 5, 2026Since v1.1.65

v1.1.66 includes 7 updates since v1.1.65. This release adds MCP setup reminders, policy downloads via API/MCP, native MSI signer evidence, and new intel block rules, along with more reliable intelligence source removal.

Highlights

  • Admins get optional MCP setup reminders with snooze and opt-out.
  • Download generated WDAC and AppLocker policies via API and hosted MCP.
  • View native MSI signer evidence in event details.
  • New ClickFix and TRegSvr block rules added to Intel Blocklist.
  • Removing multiple intelligence sources is now atomic and safe to retry.
  • Microsoft Copilot Studio MCP setup guidance added.

Fixes

  • Reliable removal of multiple intelligence sources: You can now remove multiple intelligence sources in one go without partial progress or version-change errors. The removal is atomic and safe to retry, so concurrent edits are preserved and you can retry after a failure without losing work.

Features

  • MCP setup reminders on the dashboard: Admins without a configured customer API key now see optional reminders to set up MCP, with snooze and opt-out options. The reminders appear only when eligible and can be dismissed permanently.
  • Download WDAC and AppLocker policies via API and MCP: You can now download generated WDAC and AppLocker policies through the Customer API and hosted MCP, with strict format selection and metadata like version and hash. This gives agentless customers a programmatic way to retrieve policies.
  • View native MSI signer evidence in event details: Full Details and Event Review now include a closed-by-default Signer details expander showing Windows-reported signer names for MSI events. This helps you verify the signer of blocked or audited MSI files without leaving the event view.
  • Microsoft Copilot Studio MCP setup guidance: The MCP setup guide now includes Microsoft Copilot Studio, with instructions for using the hosted connector URL and OAuth 2.0 dynamic discovery. This helps you connect Copilot Studio to MagicSword MCP.
  • Microsoft Copilot Studio MCP setup guidance: The MCP setup guide now includes Microsoft Copilot Studio, with instructions for using the hosted connector URL and OAuth 2.0 dynamic discovery. This helps you connect Copilot Studio to MagicSword MCP.
  • New ClickFix and TRegSvr block rules: Adds three new MagicSword Intel Blocklist entries: a path rule blocking PE execution from the user-profile Themes staging location, a filename rule for TREGSVR.EXE, and a hash rule for a specific TRegSvr build. These help block ClickFix-related threats while allowing legitimate use of the registration utility.
  • Windows agent 1.0.0.972 — richer MSI signer context. Correlates AppLocker Event ID 8038 with matching 8028/8029 MSI audit/block events. Portal event details and CSV exports show available Windows-reported publisher, issuer and signature evidence. Missing or ambiguous evidence preserves the original decision.
  • MagicSword MCP 0.1.7 — policy downloads and Copilot Studio guidance. Adds download_policy for complete generated Windows WDAC XML ZIPs or AppLocker XML, saved locally without installing or assigning policies. Requires policies:read and Enterprise agentless export access. Includes hosted Copilot Studio setup guidance; tenant interoperability remains unverified.
November 3, 2026Since v1.1.64

v1.1.65 includes 1 update since v1.1.64. Endpoint group pagination now returns an empty list instead of a server error when the requested page is out of range.

Highlights

  • Endpoint group pages beyond the last one now load as empty lists instead of failing.

Fixes

  • Endpoint group pages beyond the available results now return an empty list: Previously, requesting a page of endpoint groups beyond the available results (for example, an offset equal to the total count) caused a server error. Now you get an empty list, so pagination works reliably even when you reach the end.
November 3, 2026Since v1.1.63

v1.1.64 includes 11 updates since v1.1.63. This release adds policy cloning and assignment with preview, improves intel safety and Windows audit accuracy, and includes security fixes.

Highlights

  • Clone customer policies
  • Assign policies to computers or groups
  • Prevent unsafe intel selectors
  • Distinguish Windows audit events
  • Secure uninstall status reporting
  • Capture campaign attribution without cookie consent
  • Expose intelligence source timestamps
  • Scope release preparation to active endpoints

Features

  • Clone customer policies: You can now clone an existing policy to create an independent copy with all settings, rules, and intelligence links preserved, while leaving the original unchanged.
  • Capture campaign attribution without cookie consent: Campaign UTMs are now captured on arrival and delivered with signups and lead forms even if visitors ignore or reject the cookie prompt, so you don't lose attribution data.
  • Promote Slack and Attacker Tools after community signup: After subscribing to the community or blog, you'll see a dismissible Slack invitation that highlights Attacker Tools, Intel Drops, and team conversations, with a simple request flow.
  • Expose intelligence source timestamps: The intel sources API now returns created_at, updated_at, last_sync, and last_checked_at, so you can see when each source was last updated and checked.
  • Add MCP landing page to footer: The MCP landing page is now linked in the website footer under Resources, making it easier to find.

Security

  • Prevent unsafe intel selectors: Intel ingestion now rejects overly broad filename/path selectors that could create global rules, preventing accidental over-blocking and preserving detection scope.
  • Add actively exploited signers: Newly discovered actively exploited code signing certificates have been added to intelligence, helping you block threats sooner.
  • Secure uninstall status reporting: The uninstall status endpoint now requires proper authorization and returns 404 for unknown endpoints, preventing spoofing and false acknowledgments.

Fixes

  • Distinguish Windows audit events: Genuine Windows audit decisions are now correctly labeled as audit events instead of observed, and modern WDAC MSI/script records are collected, giving you more accurate event visibility.
  • Fix meta descriptions: Shortened meta descriptions on seven pages to meet SEO best practices, improving search result display.

Performance

  • Scope release preparation to active endpoints: Release preparation now skips Windows endpoints inactive for more than 30 days, focusing on active endpoints and speeding up deployments.
October 29, 2026Since v1.1.62

v1.1.63 includes 1 update since v1.1.62. Fixed WDAC policy XML generation so Boot Audit on Failure policies pass Windows ConfigCI validation.

Highlights

  • WDAC policies with Boot Audit on Failure enabled now generate schema-correct XML that Windows ConfigCI accepts.

Fixes

  • Fix Boot Audit on Failure spelling in WDAC XML: You can now generate WDAC policy XML with Boot Audit on Failure enabled that Windows ConfigCI accepts. The option is normalized to the schema-correct spelling while your saved settings and enabled/disabled state are preserved.
October 29, 2026Since v1.1.61

Free plans can now use scoped API keys and MCP, including empty audit-policy creation. This release also improves policy review, restores paid Investigate retention, and refreshes customer stories and threat intelligence.

Highlights

  • Scoped customer API keys and MCP are now available on Free.
  • Create empty audit policies through API and MCP without assigning endpoints.
  • Recover policy reviews more easily and regain subscription-backed 30-day Investigate retention.

Features

  • Create empty audit policies through API and MCP: Create an unassigned policy for Windows, macOS, or Linux, then add rules with existing tools. Strict mode remains Enterprise-only for Windows and macOS.
  • Use API keys and MCP on Free: Free organizations can use scoped customer API keys and MCP. Paid policy features and guided enforcement readiness retain their Enterprise requirements.

Improvements

  • Qualify managed service provider demo requests: Demo requests now collect MSP status and client-organization ranges for more relevant follow-up.
  • Refresh community guest thumbnails: Guest video thumbnails use consistent monochrome styling while preserving links and hover behavior.
  • Simplify anonymous customer stories: Publish the managed-services story without company identification and simplify story navigation and layouts.
  • Publish the latest Living Off the Land edition: Add the October 22–29 weekly intelligence edition and contributor ledger.
  • Align data collection dialogs: Match Collect Data and Defender collection controls to the portal design system.
  • Refresh customer story pages: Redesign customer stories and add an anonymous managed-services case study.
  • Simplify MCP connection setup: Clarify the connection sequence for configuring the hosted MCP integration.
  • Simplify the default homepage: Use the standard homepage by default, with the particle experience available through the sword Easter egg.
  • Update MITRE ATT&CK references: Refresh the catalog and technique references to MITRE ATT&CK v20.0.
  • Improve page metadata and structured data: Refine marketing page titles, descriptions, and structured data.

Operations

  • Keep database migration history consistent: Restore the source receipt for an already-applied database migration without replaying it.
  • Publish changelogs after production releases: Automatically publish the release changelog after a successful tagged portal deployment.

Security

  • Refine actively exploited signer intelligence: Update threat intelligence while excluding unsupported new certificate-wide blocks.
  • Gate macOS ClickFix rules on agent support: Repair the macOS ClickFix catalog and require compatible agent capability for shell-pipe rules. Protection requires a compatible agent and subsequent policy publication.
  • Reject unscoped wildcard intelligence rules: Prevent broad wildcard rules from entering policies through LOLRMM imports.

Fixes

  • Make policy review easier to recover: Improve retry behavior, update navigation, and exception handling in policy review and Investigate.
  • Restore paid Investigate retention: Restore subscription-backed 30-day Investigate access and show retention consistently in billing and analytics.
  • Redirect the former Premium page: Permanently redirect /premium to /plan.
October 27, 2026Since v1.1.60

v1.1.61 includes 1 update since v1.1.60. This release introduces a dedicated application control landing page with real weekly catalog figures, refreshed ThreatLocker/WDAC/AppLocker comparisons, and clearer research-to-policy messaging, while retiring the unused RMM abuse campaign page.

Highlights

  • New application control landing page with Join Free CTAs and a free 100-endpoint offer.
  • Comparisons with ThreatLocker, WDAC, and AppLocker updated to reflect current guidance.
  • Real weekly LOLRMM and LOLDrivers catalog growth figures now shown.
  • Research-to-policy messaging clarifies how MagicSword turns LOLDrivers/LOLRMM research into policies.
  • Signup links preserve ad UTMs and newsletter source for accurate attribution.
  • Retired unused RMM abuse campaign page now returns 404.
  • Growth captions now derive from snapshot data in UTC with clearer section labels.
  • Video dialog keyboard focus improved.

Features

  • New application control landing page: You can now explore MagicSword's application control offering on a dedicated page, with Join Free CTAs, a free 100-endpoint offer, and a two-hour policy-update message.
October 26, 2026Since v1.1.59

v1.1.60 includes 7 updates since v1.1.59.

Highlights

  • Release portal 1.1.60 with reviewed fixes
  • Explain built-in intelligence allowances across policy rules
  • Add newsletter landing page and Live guest highlights

Improvements

  • Release portal 1.1.60 with reviewed fixes
  • Explain built-in intelligence allowances across policy rules
  • Add newsletter landing page and Live guest highlights
  • Make MagicSword MCP easy to install and connect
  • Remove endpoint AMSI integrity card and dead display code
  • Allow common Linux system tools in the GTFOBins feed
Page 1 of 8 · 77 releases