davinci-resolve-mcp 2.87.0 → 2.87.2

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.
package/AGENTS.md CHANGED
@@ -73,7 +73,13 @@ semantics.
73
73
  `docs/reference/api-limitations.md` — when you document a new Resolve API
74
74
  limitation, add a `submit`-tagged entry to `src/utils/api_truth.py` and
75
75
  regenerate with `scripts/gen_api_limitations.py` (a drift guard enforces it)
76
- - Bundled Resolve API text: `docs/reference/resolve_scripting_api.txt`
76
+ - Bundled Resolve API text: `docs/reference/resolve_scripting_api.txt` — refresh
77
+ it by copying the shipped `Developer/Scripting/README.txt` over it wholesale,
78
+ never by hand-adding the lines you already know about. The file's only value
79
+ is that a diff against it is trustworthy; a hand-patch carries a newer
80
+ `Last Updated:` header while still hiding everything you did not know to look
81
+ for. That is exactly how it sat eight weeks stale at 26 May 2026 while ten
82
+ documented 21.0.4 methods went unwired (PR #139)
77
83
 
78
84
  ## Key Paths
79
85
 
package/CHANGELOG.md CHANGED
@@ -2,6 +2,71 @@
2
2
 
3
3
  Release history for the DaVinci Resolve MCP Server. The latest release is summarized in the root README; older entries live here to keep the README focused.
4
4
 
5
+ ## What's New in v2.87.2
6
+
7
+ A refused `SetSetting` now says why, when the ledger already knows. Issue #141,
8
+ reported by @jus-kim.
9
+
10
+ ### Fixed
11
+
12
+ - **`project_settings set_setting` returned a bare `{"success": false}` for a
13
+ key that can never be written.** `Project.SetSetting('timelinePlaybackFrameRate')`
14
+ refuses every value form, before and after a timeline exists — measured in
15
+ PR #99, written into `api_truth`, published in `api-limitations.md`, and
16
+ invisible at the one moment it mattered. A bare `false` reads as *your value
17
+ was wrong*, which sends a caller into retrying string, int, and float for a
18
+ key with no writable path at all. A refusal now carries the ledger entry for
19
+ that key: what is really happening, and the UI step that is the way through.
20
+ `timeline set_setting` gets the same treatment.
21
+ - The match is deliberately narrow. It requires the exact quoted key on the
22
+ right object — `Project.SetSetting('x')` will not be handed to a `Timeline`
23
+ refusal, and a substring like `timeline` will not collect the
24
+ `timelinePlaybackFrameRate` entry. An unmeasured refusal stays bare, because
25
+ inventing an explanation for a failure nobody measured is the thing this
26
+ ledger exists to prevent. The write is always attempted first, so a key that
27
+ starts working in a later build reports plain success.
28
+
29
+ ### Documentation
30
+
31
+ - **The `timelinePlaybackFrameRate` ledger entry carries the second report.**
32
+ Issue #141 confirms it independently on **Resolve 20.2**, against a freshly
33
+ created project whose timeline rate already read 60 — so a matching
34
+ `timelineFrameRate` does not unlock the write, which the PR #99 measurement
35
+ alone left open. The reporter's workaround is now recorded too: for repeat
36
+ setups, duplicate a project that already carries the wanted playback rate
37
+ rather than creating one and trying to write it.
38
+
39
+ ## What's New in v2.87.1
40
+
41
+ Follow-up evidence from @legionsound on PR #139, plus the process fix for the
42
+ tooling failure that PR exposed.
43
+
44
+ ### Changed
45
+
46
+ - **`import_user_preferences_preset` now says what the preset ended up called.**
47
+ The no-name path was the one branch in v2.87.0 with nothing behind it — it
48
+ calls the single-argument binding, which would have been a `TypeError` rather
49
+ than a graceful failure if the binding wanted two positionals. Round-tripped
50
+ on Studio 21.0.4.5 (export → delete → import both ways): the single-arg form
51
+ returns `True`, and the imported preset is **named after the file**. The
52
+ answer now carries that, so a caller who passed no name knows to read the
53
+ name back with `list_user_preferences_presets` instead of guessing.
54
+
55
+ ### Documentation
56
+
57
+ - **`api-coverage.md`: the import/export rows carry the round-trip**, not
58
+ `dir()` membership. Still 🔬 and still 23 untested of 361 — the counting
59
+ convention is unchanged, because none of this was executed here. What changed
60
+ is the strength of the contributor's evidence behind two of the rows.
61
+ - **`AGENTS.md` now states how the API snapshot must be refreshed:** copy the
62
+ shipped `Developer/Scripting/README.txt` over it wholesale, never hand-add the
63
+ lines you already know about. A hand-patch carries a newer `Last Updated:`
64
+ header while hiding everything you did not know to look for — which is
65
+ precisely how the file sat eight weeks stale at 26 May 2026 while ten
66
+ documented 21.0.4 methods went unwired. The v2.87.0 snapshot was verified as a
67
+ wholesale replace (md5 `d732b3f6c1da08dc516bc8f80c5acd92`, byte-identical to
68
+ the 21.0.4.5 shipped file per the contributor).
69
+
5
70
  ## What's New in v2.87.0
6
71
 
7
72
  The ten Resolve 21.0.4 scripting surfaces that the 24 Jul 2026 README refresh
package/README.md CHANGED
@@ -2,7 +2,7 @@
2
2
 
3
3
  English | [简体中文](README.zh-CN.md)
4
4
 
5
- [![Version](https://img.shields.io/badge/version-2.87.0-blue.svg)](https://github.com/samuelgursky/davinci-resolve-mcp/releases)
5
+ [![Version](https://img.shields.io/badge/version-2.87.2-blue.svg)](https://github.com/samuelgursky/davinci-resolve-mcp/releases)
6
6
  [![npm](https://img.shields.io/npm/v/davinci-resolve-mcp.svg?label=npm&color=CB3837)](https://www.npmjs.com/package/davinci-resolve-mcp)
7
7
  [![API Coverage](https://img.shields.io/badge/API%20Coverage-100%25-brightgreen.svg)](docs/reference/api-coverage.md)
8
8
  [![Tools](https://img.shields.io/badge/MCP%20Tools-34%20(341%20full)-blue.svg)](#server-modes)
package/README.zh-CN.md CHANGED
@@ -2,7 +2,7 @@
2
2
 
3
3
  [English](README.md) | 简体中文
4
4
 
5
- [![Version](https://img.shields.io/badge/version-2.87.0-blue.svg)](https://github.com/samuelgursky/davinci-resolve-mcp/releases)
5
+ [![Version](https://img.shields.io/badge/version-2.87.2-blue.svg)](https://github.com/samuelgursky/davinci-resolve-mcp/releases)
6
6
  [![npm](https://img.shields.io/npm/v/davinci-resolve-mcp.svg?label=npm&color=CB3837)](https://www.npmjs.com/package/davinci-resolve-mcp)
7
7
  [![API Coverage](https://img.shields.io/badge/API%20Coverage-100%25-brightgreen.svg)](docs/reference/api-coverage.md)
8
8
  [![Tools](https://img.shields.io/badge/MCP%20Tools-34%20(341%20full)-blue.svg)](#服务器模式)
@@ -12,7 +12,7 @@
12
12
  [![Python](https://img.shields.io/badge/python-3.10+-green.svg)](https://www.python.org/downloads/)
13
13
  [![License](https://img.shields.io/badge/license-MIT-blue.svg)](https://opensource.org/licenses/MIT)
14
14
 
15
- > 本翻译对应 v2.87.0 版 README。如与英文原版有出入,以 [英文原版](README.md) 为准。
15
+ > 本翻译对应 v2.87.2 版 README。如与英文原版有出入,以 [英文原版](README.md) 为准。
16
16
 
17
17
  一个 Model Context Protocol (MCP) 服务器,让 AI 助手通过官方脚本 API 控制 DaVinci Resolve Studio(达芬奇)。它提供完整的 API 覆盖,外加带护栏的工作流助手,涵盖剪辑、媒体池整理、渲染设置、审阅标记、调色、Fusion、Fairlight、项目生命周期任务、扩展开发,以及不碰源媒体的媒体分析。
18
18
 
@@ -157,7 +157,7 @@ checkable rather than asserted.
157
157
  | `Resolve.SaveUserPreferencesPreset` | Requires a Resolve 21.0.4 build; the validation machine runs Studio 19.1.3 and free 21.0.3 | Yes |
158
158
  | `Resolve.LoadUserPreferencesPreset` | Deliberately not executed even by the reporter: swaps the user's global preferences session-wide | No |
159
159
  | `Resolve.DeleteUserPreferencesPreset` | Requires a Resolve 21.0.4 build; the validation machine runs Studio 19.1.3 and free 21.0.3 | Yes |
160
- | `Resolve.ImportUserPreferencesPreset` | Requires a Resolve 21.0.4 build and a preset file to import | Yes |
160
+ | `Resolve.ImportUserPreferencesPreset` | Requires a Resolve 21.0.4 build; round-tripped by the reporter on 21.0.4.5 in both call shapes (PR #139), not here | Yes |
161
161
  | `Resolve.ExportUserPreferencesPreset` | Requires a Resolve 21.0.4 build; the validation machine runs Studio 19.1.3 and free 21.0.3 | Yes |
162
162
  | `PM.GetProjectAttributesInCurrentFolder` | Requires a Resolve 21.0.4 build; the validation machine runs Studio 19.1.3 and free 21.0.3 | Yes |
163
163
 
@@ -218,8 +218,8 @@ Every method in the DaVinci Resolve Scripting API and its test status. Methods a
218
218
  | 28 | `SaveUserPreferencesPreset(presetName)` | 🔬 | Resolve 21.0.4. Not executed here; part of the issue #138 round-trip on 21.0.4.5 |
219
219
  | 29 | `LoadUserPreferencesPreset(presetName)` | 🔬 | Resolve 21.0.4. `dir()`-present on 21.0.4.5; **not executed** even by the reporter — swaps the user's global preferences session-wide (issue #138) |
220
220
  | 30 | `DeleteUserPreferencesPreset(presetName)` | 🔬 | Resolve 21.0.4. Not executed here; part of the issue #138 round-trip on 21.0.4.5 |
221
- | 31 | `ImportUserPreferencesPreset(presetFilePath, presetName)` | 🔬 | Resolve 21.0.4. `dir()`-present on 21.0.4.5 only (issue #138) — no preset file was on hand. Per the README the imported preset is **not** auto-loaded |
222
- | 32 | `ExportUserPreferencesPreset(presetName, exportPath)` | 🔬 | Resolve 21.0.4. `dir()`-present on 21.0.4.5 only (issue #138) |
221
+ | 31 | `ImportUserPreferencesPreset(presetFilePath, presetName)` | 🔬 | Resolve 21.0.4. Not executed here; reporter's export→delete→import round-trip on 21.0.4.5 (PR #139) covers **both** call shapes single-arg returns `True` rather than a `TypeError`, and the imported preset is then named after the file; the two-arg form takes the given name. Per the README the imported preset is **not** auto-loaded, and the round-trip did not load either import |
222
+ | 32 | `ExportUserPreferencesPreset(presetName, exportPath)` | 🔬 | Resolve 21.0.4. Not executed here; reported writing the file at the given path with no extension appended, on 21.0.4.5 (PR #139) |
223
223
 
224
224
  ### ProjectManager
225
225
 
@@ -44,8 +44,8 @@ equivalent, blocking full automation.
44
44
  ### Project.SetSetting('timelinePlaybackFrameRate')
45
45
 
46
46
  - **Object:** `Project`
47
- - **Behavior:** Returns False for every value form tried (string, int, float), both before and after a timeline exists, so the playback frame rate cannot be set from the API at all. Reported by a community contributor against Resolve Studio while assembling a vertical timeline (PR #99).
48
- - **Workaround / current handling:** Ask the user to set it in Project Settings > Master Settings > Playback frame rate as a SETUP step, before any timeline exists. Read it back to confirm; do not report it as set on the strength of the call alone.
47
+ - **Behavior:** Returns False for every value form tried (string, int, float), both before and after a timeline exists, so the playback frame rate cannot be set from the API at all. Reported by a community contributor against Resolve Studio while assembling a vertical timeline (PR #99), and independently on Resolve 20.2 against a freshly created project whose timeline rate already read 60 (issue #141) — so a matching timelineFrameRate does not unlock the write.
48
+ - **Workaround / current handling:** Ask the user to set it in Project Settings > Master Settings > Playback frame rate as a SETUP step, before any timeline exists. Read it back to confirm; do not report it as set on the strength of the call alone. The issue #141 reporter's workaround is worth passing on for repeat setups: duplicate a project that already carries the wanted playback rate rather than creating one and trying to write it.
49
49
  - **Tags:** project-settings, silent-failure, timeline
50
50
 
51
51
  ### Timeline.GetCurrentClipThumbnailImage (Color page only)
package/install.py CHANGED
@@ -36,7 +36,7 @@ from src.utils.update_check import (
36
36
 
37
37
  # ─── Version ──────────────────────────────────────────────────────────────────
38
38
 
39
- VERSION = "2.87.0"
39
+ VERSION = "2.87.2"
40
40
  # Only hard floor: mcp[cli] requires Python 3.10+. There is no upper bound —
41
41
  # Resolve's scripting bridge loads into newer interpreters on recent builds
42
42
  # (Python 3.14 verified against Resolve Studio 20.3.2). Older Resolve builds
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "davinci-resolve-mcp",
3
- "version": "2.87.0",
3
+ "version": "2.87.2",
4
4
  "description": "NPM bootstrapper for the DaVinci Resolve MCP Server.",
5
5
  "license": "MIT",
6
6
  "author": "Samuel Gursky <samgursky@gmail.com>",
@@ -85,7 +85,7 @@ if not logging.getLogger().handlers:
85
85
  handlers=[logging.StreamHandler()],
86
86
  )
87
87
 
88
- VERSION = "2.87.0"
88
+ VERSION = "2.87.2"
89
89
  logger = logging.getLogger("davinci-resolve-mcp")
90
90
  logger.info(f"Starting DaVinci Resolve MCP Server v{VERSION}")
91
91
  logger.info(f"Detected platform: {get_platform()}")
package/src/server.py CHANGED
@@ -11,7 +11,7 @@ Usage:
11
11
  python src/server.py --full # Start the 341-tool granular server instead
12
12
  """
13
13
 
14
- VERSION = "2.87.0"
14
+ VERSION = "2.87.2"
15
15
 
16
16
  import base64
17
17
  import os
@@ -13589,7 +13589,10 @@ def resolve_control(action: str, params: Optional[Dict[str, Any]] = None) -> Dic
13589
13589
  delete_user_preferences_preset(name) -> {success} — Resolve 21.0.4+
13590
13590
  import_user_preferences_preset(path, name?) -> {success} — Resolve 21.0.4+.
13591
13591
  The imported preset is NOT auto-loaded; follow with
13592
- load_user_preferences_preset to activate it.
13592
+ load_user_preferences_preset to activate it. Omitting name is safe
13593
+ (reported on 21.0.4.5: single-arg returns True, not a TypeError) and
13594
+ the preset is then named after the file, so pass name when the preset
13595
+ should be called something the filename does not say.
13593
13596
  export_user_preferences_preset(name, path) -> {success} — Resolve 21.0.4+
13594
13597
  open_control_panel(port?, host?, open_browser?) -> {success, url, pid, port, status}
13595
13598
  — Launches the analysis control panel (src/analysis_dashboard.py) as a background process.
@@ -13862,8 +13865,11 @@ def resolve_control(action: str, params: Optional[Dict[str, Any]] = None) -> Dic
13862
13865
  ok = bool(r.ImportUserPreferencesPreset(p["path"], p["name"]))
13863
13866
  else:
13864
13867
  ok = bool(r.ImportUserPreferencesPreset(p["path"]))
13865
- return {"success": ok,
13866
- "note": "The imported preset is not auto-loaded; use load_user_preferences_preset to activate it."}
13868
+ note = "The imported preset is not auto-loaded; use load_user_preferences_preset to activate it."
13869
+ if not p.get("name"):
13870
+ note += (" No name was given, so the preset takes its name from the file — "
13871
+ "list_user_preferences_presets to read it back.")
13872
+ return {"success": ok, "note": note}
13867
13873
  elif action == "export_user_preferences_preset":
13868
13874
  missing = _requires_method(r, "ExportUserPreferencesPreset", "21.0.4")
13869
13875
  if missing:
@@ -16011,6 +16017,37 @@ def project_manager_database(action: str, params: Optional[Dict[str, Any]] = Non
16011
16017
  # TOOL 8: project_settings
16012
16018
  # ═══════════════════════════════════════════════════════════════════════════════
16013
16019
 
16020
+ def _setting_limitation(name: Any, obj: str = "Project") -> Optional[Dict[str, Any]]:
16021
+ """The api_truth entry for a settings key on `obj`, when one exists.
16022
+
16023
+ `SetSetting` reports a refusal as a bare `False` with no reason, and for
16024
+ several keys this repo has already measured the reason and written it down —
16025
+ `timelinePlaybackFrameRate` returns False for every value form, before and
16026
+ after a timeline exists (issue #141, PR #99). A caller who gets
16027
+ `{"success": false}` has no way to tell "you passed a bad value" from "this
16028
+ key cannot be written from the API at all", and the second one is a
16029
+ different task: it has to go to the user as a UI step.
16030
+
16031
+ Matched narrowly on purpose. The entry must name this exact key *and* be
16032
+ `obj.SetSetting`, because attaching an unrelated explanation to a failure is
16033
+ worse than attaching none — it reads as a diagnosis. `Project` and
16034
+ `Timeline` both have a `SetSetting` and their keys overlap by name, so the
16035
+ object is part of the match rather than assumed.
16036
+ """
16037
+ if not isinstance(name, str) or not name:
16038
+ return None
16039
+ prefix = f"{obj}.SetSetting"
16040
+ quoted = f"'{name}'"
16041
+ for entry in lookup_api_truth(name):
16042
+ symbol = entry.get("symbol", "")
16043
+ # The quoted form is what makes this an exact key match: `name in
16044
+ # symbol` would hand the timelinePlaybackFrameRate entry to anything
16045
+ # that is a substring of it, "timeline" included.
16046
+ if symbol.startswith(prefix) and quoted in symbol:
16047
+ return entry
16048
+ return None
16049
+
16050
+
16014
16051
  @mcp.tool()
16015
16052
  @_guard_missing_params
16016
16053
  def project_settings(action: str, params: Optional[Dict[str, Any]] = None) -> Dict[str, Any]:
@@ -16020,7 +16057,9 @@ def project_settings(action: str, params: Optional[Dict[str, Any]] = None) -> Di
16020
16057
  get_name() -> {name}
16021
16058
  set_name(name) -> {success}
16022
16059
  get_setting(name?) -> {settings} — omit name for all settings
16023
- set_setting(name, value) -> {success}
16060
+ set_setting(name, value) -> {success, known_limitation?}
16061
+ A refusal carries the api_truth entry for that key when one exists —
16062
+ several settings cannot be written from the API at all.
16024
16063
  get_unique_id() -> {id}
16025
16064
  get_presets() -> {presets}
16026
16065
  set_preset(name) -> {success}
@@ -16056,7 +16095,20 @@ def project_settings(action: str, params: Optional[Dict[str, Any]] = None) -> Di
16056
16095
  return _err("set_setting requires name")
16057
16096
  if "value" not in p:
16058
16097
  return _err("set_setting requires value")
16059
- return {"success": bool(proj.SetSetting(p["name"], p["value"]))}
16098
+ if bool(proj.SetSetting(p["name"], p["value"])):
16099
+ return {"success": True}
16100
+ known = _setting_limitation(p["name"])
16101
+ if not known:
16102
+ return {"success": False}
16103
+ return {
16104
+ "success": False,
16105
+ "known_limitation": {
16106
+ "symbol": known.get("symbol"),
16107
+ "reality": known.get("reality"),
16108
+ "recommended": known.get("recommended"),
16109
+ "ledger_verified_on": _API_TRUTH_VERIFIED_ON,
16110
+ },
16111
+ }
16060
16112
  elif action == "get_unique_id":
16061
16113
  return {"id": proj.GetUniqueId()}
16062
16114
  elif action == "get_presets":
@@ -21396,7 +21448,8 @@ def timeline(action: str, params: Optional[Dict[str, Any]] = None) -> Dict[str,
21396
21448
  export(path, type, subtype?, background?) -> {success | job_id} — type: AAF, EDL, FCPXML, etc.
21397
21449
  UNSAFE. No path sandboxing. Prefer export_timeline_checked.
21398
21450
  get_setting(name?) -> {settings}
21399
- set_setting(name, value) -> {success}
21451
+ set_setting(name, value) -> {success, known_limitation?}
21452
+ A refusal carries the api_truth entry for that key when one exists.
21400
21453
  insert_generator(name) -> {success}
21401
21454
  insert_fusion_generator(name) -> {success}
21402
21455
  insert_fusion_composition() -> {success}
@@ -21756,7 +21809,20 @@ def timeline(action: str, params: Optional[Dict[str, Any]] = None) -> Dict[str,
21756
21809
  elif action == "get_setting":
21757
21810
  return {"settings": _ser(tl.GetSetting(p.get("name", "")))}
21758
21811
  elif action == "set_setting":
21759
- return {"success": bool(tl.SetSetting(p["name"], p["value"]))}
21812
+ if bool(tl.SetSetting(p["name"], p["value"])):
21813
+ return {"success": True}
21814
+ known = _setting_limitation(p["name"], obj="Timeline")
21815
+ if not known:
21816
+ return {"success": False}
21817
+ return {
21818
+ "success": False,
21819
+ "known_limitation": {
21820
+ "symbol": known.get("symbol"),
21821
+ "reality": known.get("reality"),
21822
+ "recommended": known.get("recommended"),
21823
+ "ledger_verified_on": _API_TRUTH_VERIFIED_ON,
21824
+ },
21825
+ }
21760
21826
  elif action == "insert_generator":
21761
21827
  r = tl.InsertGeneratorIntoTimeline(p["name"])
21762
21828
  return _ok() if r else _err("Failed to insert generator")
@@ -124,11 +124,18 @@ API_TRUTH: List[Dict[str, Any]] = [
124
124
  "float), both before and after a timeline exists, so the "
125
125
  "playback frame rate cannot be set from the API at all. "
126
126
  "Reported by a community contributor against Resolve Studio "
127
- "while assembling a vertical timeline (PR #99).",
127
+ "while assembling a vertical timeline (PR #99), and "
128
+ "independently on Resolve 20.2 against a freshly created "
129
+ "project whose timeline rate already read 60 (issue #141) — "
130
+ "so a matching timelineFrameRate does not unlock the write.",
128
131
  "recommended": "Ask the user to set it in Project Settings > Master "
129
132
  "Settings > Playback frame rate as a SETUP step, before "
130
133
  "any timeline exists. Read it back to confirm; do not "
131
- "report it as set on the strength of the call alone.",
134
+ "report it as set on the strength of the call alone. The "
135
+ "issue #141 reporter's workaround is worth passing on "
136
+ "for repeat setups: duplicate a project that already "
137
+ "carries the wanted playback rate rather than creating "
138
+ "one and trying to write it.",
132
139
  "tags": ["project-settings", "silent-failure", "timeline"],
133
140
  "submit": "missing",
134
141
  },