dodo-mcp 1.4.1 → 1.4.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1,5 +1,50 @@
1
1
  # DODO MCP — Test Report
2
2
 
3
+ ## 1.4.2 Recovery best-effort patch — Windows, 2026-11-11
4
+
5
+ Focused checks on Windows / Node 24.22.1: **build, typecheck and lint passed**.
6
+ These results cover the Recovery patch and are not a full native release gate.
7
+
8
+ | Check | Passed | Failed | Skipped |
9
+ |---|---:|---:|---:|
10
+ | Main Recovery regressions (7 files) | 69 | 0 | 2 |
11
+ | Recovery-only storage failure at boot and pinned directory identity | 2 | 0 | 0 |
12
+ | Owner UI in installed system Chrome | 1 | 0 | 0 |
13
+ | Fresh tarball install and packaging | 17 | 0 | 0 |
14
+ | Total focused checks | 89 | 0 | 2 |
15
+
16
+ The main group's skips were the Windows directory-fsync case and the existing
17
+ owner UI case whose bundled Chromium was unavailable. A separate UI regression
18
+ used installed system Chrome and exercised the actual Recovery dashboard code.
19
+
20
+ Real isolated SQLite/workspace/process fixtures verify persisted secret skipping
21
+ with an AWS-like test key after restart; actual `echo ok` and edits despite
22
+ legacy strict/Git settings or unavailable storage; corrupt prior snapshots and
23
+ policy/drift metadata; relative failure paths and stale-path clearing; live
24
+ authority revocation; and failed required backups before a reviewed restore.
25
+ Bookkeeping trigger failures preserve the operation's result and do not repeat
26
+ effects. Existing regressions retain stale-hash, secret/path, backup integrity
27
+ and actual partial-journal recovery checks.
28
+
29
+ The storage restart test replaces only a disposable Recovery directory with a
30
+ plain file, confirms the server boots and performs normal work, and verifies that
31
+ the occupied file and old backup bytes stay intact. A separate identity test
32
+ rejects a replacement directory after successful lazy initialization.
33
+
34
+ The Chrome case checks legacy `BLOCKED` diagnostics, secret-skip settings in both
35
+ states, project-only skip actions and an unsuccessful retry notice. Packaging
36
+ checks install a fresh tarball and verify its Compact MCP backup/restore flow and
37
+ the new overview fields.
38
+
39
+ **Not completed:** the full Windows suite/native gate (the host lacks enabled
40
+ Administrator and symlink prerequisites), native Linux/macOS validation, and live
41
+ gufile/production acceptance. The earlier production-only npm audit reported
42
+ 10 affected packages (1 critical, 5 high, 4 moderate); this Recovery patch does not
43
+ remediate those advisories or claim a clean dependency audit. Publication is
44
+ verified separately from these tests.
45
+
46
+ See [release scope and upgrade steps](RELEASE_1.4.2.md).
47
+
3
48
  ## 1.3.2 Recovery maintenance — macOS release gate, 2026-09-25
4
49
 
5
50
  `AUTOMATED_PASS` on macOS: build, typecheck and lint; full core suite **926 passed,
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "dodo-mcp",
3
- "version": "1.4.1",
3
+ "version": "1.4.2",
4
4
  "author": "Arthittakun",
5
5
  "description": "DODO — local-first MCP server with multi-project coding, AI providers, sub-agents, context, memory, media and workflow tools; embedded OAuth and workspace permissions.",
6
6
  "type": "module",
@@ -56,6 +56,7 @@
56
56
  "docs/RELEASE_1.3.2.md",
57
57
  "docs/RELEASE_1.4.0.md",
58
58
  "docs/RELEASE_1.4.1.md",
59
+ "docs/RELEASE_1.4.2.md",
59
60
  "docs/RELEASE_NOTES.md",
60
61
  "docs/SCHEDULES.md",
61
62
  "docs/SECURITY.md",