dsh-vibe-math 2.6.0 → 2.7.0
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/README.en.md +156 -389
- package/README.md +156 -390
- package/cordis.patch.yml +56 -14
- package/docs/AUDIT-CHECKLIST.md +2 -2
- package/docs/final-paper.md +1 -0
- package/docs/formal-verification.md +184 -29
- package/docs/math-computation.md +121 -0
- package/docs/release-notes/RELEASE-NOTES-2.7.0.en.md +151 -0
- package/docs/release-notes/RELEASE-NOTES-2.7.0.md +115 -0
- package/docs/test-timing.md +24 -12
- package/installer.js +7 -4
- package/package.json +18 -2
- package/prompt-corpus-persona/persona-corpus.json +12 -12
- package/prompt-corpus-persona/persona-corpus.md +116 -32
- package/prompt-corpus-v2/formal-verify-v2.json +116 -92
- package/prompt-corpus-v2/formal-verify-v2.md +1430 -55
- package/prompt-corpus-v3/formal-verify-v3.json +101 -53
- package/prompt-corpus-v3/formal-verify-v3.md +1001 -22
- package/prompt-corpus-v4/formal-verify-v4.json +14 -14
- package/prompt-corpus-v4/formal-verify-v4.md +247 -14
- package/prompt-corpus-v5/prompt-corpus-v5.json +62 -62
- package/prompt-corpus-v5/prompt-corpus-v5.md +570 -3
- package/tests/audit-math-computation-contract.mjs +191 -0
- package/tests/audit-math-computation-parity.mjs +169 -0
- package/tests/audit-math-computation-sensitivity.mjs +231 -0
- package/tests/audit-persona-sensitivity.mjs +4 -4
- package/tests/audit-persona-surface.test.mjs +38 -10
- package/tests/audit-v5-integrity.mjs +14 -0
- package/tests/e2e-v5-round2.test.mjs +452 -2
- package/tests/formal-verify-v2.test.mjs +292 -6
- package/tests/formal-verify-v3.test.mjs +271 -4
- package/tests/formal-verify-v4.test.mjs +295 -6
- package/tests/formal-verify-v5.test.mjs +9 -1
- package/tests/helpers/math-computation-fake-seam.mjs +135 -0
- package/tests/math-computation-shared.test.mjs +504 -0
- package/tests/prompt-v5-integrity.test.mjs +5 -5
- package/vibe-math-v2/agent.cordis.yml +20 -8
- package/vibe-math-v2/math-computation.js +936 -0
- package/vibe-math-v2/math-engines.js +156 -0
- package/vibe-math-v2/vibe-math-v2.js +837 -40
- package/vibe-math-v2//345/256/236/347/216/260/346/226/271/346/241/210.md +6 -0
- package/vibe-math-v3/agent.cordis.yml +20 -8
- package/vibe-math-v3/math-computation.js +936 -0
- package/vibe-math-v3/math-engines.js +156 -0
- package/vibe-math-v3/vibe-math-v3.js +828 -47
- package/vibe-math-v3//345/256/236/347/216/260/346/226/271/346/241/210.md +6 -0
- package/vibe-math-v4/agent.cordis.yml +34 -6
- package/vibe-math-v4/math-computation.js +936 -0
- package/vibe-math-v4/math-engines.js +156 -0
- package/vibe-math-v4/vibe-math-v4.js +801 -31
- package/vibe-math-v4//345/256/236/347/216/260/346/226/271/346/241/210.md +6 -0
- package/vibe-math-v5/agent.cordis.yml +38 -6
- package/vibe-math-v5/math-computation.js +936 -0
- package/vibe-math-v5/math-engines.js +156 -0
- package/vibe-math-v5/vibe-math-v5.js +1120 -37
- package/vibe-math-v5//345/256/236/347/216/260/346/226/271/346/241/210.md +83 -3
- package/vibe-math-v5//346/236/266/346/236/204/345/233/276.md +251 -0
|
@@ -5,7 +5,7 @@
|
|
|
5
5
|
|
|
6
6
|
## vibe-math-v2
|
|
7
7
|
|
|
8
|
-
- 注册工具数:**
|
|
8
|
+
- 注册工具数:**28**
|
|
9
9
|
- 斜杠命令 hint:`start|resume|pause|abort|status|report|mode <auto|manual>|setup|save|template [global|project]|add <id> <desc>|add-proposition <id> <概述>|list-propositions|project [list|new <name>|<name>]|decisions|agents|paper [lang=zh|en] [format=both|md|tex] [force]`
|
|
10
10
|
|
|
11
11
|
### config.prefix
|
|
@@ -33,9 +33,15 @@ programmatically runs explorer (direction setting) → per-direction solvers (ag
|
|
|
33
33
|
- vibe_math_new_project / vibe_math_set_project / vibe_math_list_projects — per-project folders.
|
|
34
34
|
- vibe_math_list_decisions / vibe_math_decide {id, action: approve|reject|override, verdict?} — resolve manual decisions.
|
|
35
35
|
- vibe_math_list_agents / vibe_math_message_agent / vibe_math_interrupt_agent — inspect / steer / interrupt subagents.
|
|
36
|
-
- vibe_math_lean_run / vibe_math_lean_archive / vibe_math_lean_lib
|
|
37
|
-
verification (execute / archive / list the reuse
|
|
38
|
-
|
|
36
|
+
- vibe_math_lean_run / vibe_math_lean_archive / vibe_math_lean_lib / vibe_math_lean_read /
|
|
37
|
+
vibe_math_lean_job — Lean formal verification (execute / archive / list + read the reuse
|
|
38
|
+
library / inspect or wait for a background compile job). leanAsync=true (default) queues the
|
|
39
|
+
compile and returns a jobId; nothing counts as proved until that job settles ok (so "有把握"
|
|
40
|
+
is not the same as "已验证"). The scheduler's child agents use them too; they work in every
|
|
41
|
+
mode.
|
|
42
|
+
|
|
43
|
+
- math_computation {op: probe|run|receipt|install, engine, mode: code|file|expr, …} — 先 probe 预检引擎/包/许可,再调引擎计算并把脚本与输出归档成可复核回执;缺引擎/缺包只报告与给安装指引/计划;shell 兜底不算归档。
|
|
44
|
+
- 归档→编辑→重跑:mode:'code' 的脚本原件在回执的 scriptPath(Computation/<id>/script.<ext>),你可以用普通文件工具打开并编辑它;编辑后用 mode:'file' 指向它重跑,会写出一份**新回执/新 attempt**(含新的 scriptHash)。**旧回执对修改后的代码无效**——报告里必须引用与当前代码哈希一致的那份回执;工具会在 scriptChanged / scriptChangedDuringRun 为 true 时显式告警。
|
|
39
45
|
|
|
40
46
|
Final paper: strict termination writes Paper/<project>/{paper.md,paper.tex,paper.pdf,paper.meta.json,paper.log.md}
|
|
41
47
|
automatically — a dedicated paper-writing child assembles the solved material (finalPaper=false disables only that
|
|
@@ -74,7 +80,7 @@ LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
|
74
80
|
explicit, reasoned blocker record; the scheduler is never wedged by it).
|
|
75
81
|
- Paths: work file Formal/<id>.lean; archived proof Verified/Lean/<id>.lean; reusable
|
|
76
82
|
definitions VibeMath/Formal/Lib/; proved lemmas VibeMath/Formal/Proved/.
|
|
77
|
-
- The toolchain knobs leanCommand / leanArgs / leanTimeoutMs are tunable as well
|
|
83
|
+
- The toolchain knobs leanCommand / leanArgs / leanTimeoutMs / leanAsync / leanInitiative / leanSearchPaths / leanJobsMaxParallel are tunable as well
|
|
78
84
|
(e.g. leanCommand='lake' with leanArgs=['env','lean']); a missing Lean binary is
|
|
79
85
|
reported as LEAN_NOT_FOUND and still lets the code be written and archived.
|
|
80
86
|
- vibe_math_status / vibe_math_report show the mode, the per-object formal status and the
|
|
@@ -107,9 +113,15 @@ programmatically runs explorer (direction setting) → per-direction solvers (ag
|
|
|
107
113
|
- vibe_math_new_project / vibe_math_set_project / vibe_math_list_projects — per-project folders.
|
|
108
114
|
- vibe_math_list_decisions / vibe_math_decide {id, action: approve|reject|override, verdict?} — resolve manual decisions.
|
|
109
115
|
- vibe_math_list_agents / vibe_math_message_agent / vibe_math_interrupt_agent — inspect / steer / interrupt subagents.
|
|
110
|
-
- vibe_math_lean_run / vibe_math_lean_archive / vibe_math_lean_lib
|
|
111
|
-
verification (execute / archive / list the reuse
|
|
112
|
-
|
|
116
|
+
- vibe_math_lean_run / vibe_math_lean_archive / vibe_math_lean_lib / vibe_math_lean_read /
|
|
117
|
+
vibe_math_lean_job — Lean formal verification (execute / archive / list + read the reuse
|
|
118
|
+
library / inspect or wait for a background compile job). leanAsync=true (default) queues the
|
|
119
|
+
compile and returns a jobId; nothing counts as proved until that job settles ok (so "有把握"
|
|
120
|
+
is not the same as "已验证"). The scheduler's child agents use them too; they work in every
|
|
121
|
+
mode.
|
|
122
|
+
|
|
123
|
+
- math_computation {op: probe|run|receipt|install, engine, mode: code|file|expr, …} — 先 probe 预检引擎/包/许可,再调引擎计算并把脚本与输出归档成可复核回执;缺引擎/缺包只报告与给安装指引/计划;shell 兜底不算归档。
|
|
124
|
+
- 归档→编辑→重跑:mode:'code' 的脚本原件在回执的 scriptPath(Computation/<id>/script.<ext>),你可以用普通文件工具打开并编辑它;编辑后用 mode:'file' 指向它重跑,会写出一份**新回执/新 attempt**(含新的 scriptHash)。**旧回执对修改后的代码无效**——报告里必须引用与当前代码哈希一致的那份回执;工具会在 scriptChanged / scriptChangedDuringRun 为 true 时显式告警。
|
|
113
125
|
|
|
114
126
|
Final paper: strict termination writes Paper/<project>/{paper.md,paper.tex,paper.pdf,paper.meta.json,paper.log.md}
|
|
115
127
|
automatically — a dedicated paper-writing child assembles the solved material (finalPaper=false disables only that
|
|
@@ -148,7 +160,7 @@ LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
|
148
160
|
explicit, reasoned blocker record; the scheduler is never wedged by it).
|
|
149
161
|
- Paths: work file Formal/<id>.lean; archived proof Verified/Lean/<id>.lean; reusable
|
|
150
162
|
definitions VibeMath/Formal/Lib/; proved lemmas VibeMath/Formal/Proved/.
|
|
151
|
-
- The toolchain knobs leanCommand / leanArgs / leanTimeoutMs are tunable as well
|
|
163
|
+
- The toolchain knobs leanCommand / leanArgs / leanTimeoutMs / leanAsync / leanInitiative / leanSearchPaths / leanJobsMaxParallel are tunable as well
|
|
152
164
|
(e.g. leanCommand='lake' with leanArgs=['env','lean']); a missing Lean binary is
|
|
153
165
|
reported as LEAN_NOT_FOUND and still lets the code be written and archived.
|
|
154
166
|
- vibe_math_status / vibe_math_report show the mode, the per-object formal status and the
|
|
@@ -159,7 +171,7 @@ LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
|
159
171
|
|
|
160
172
|
## vibe-math-v3
|
|
161
173
|
|
|
162
|
-
- 注册工具数:**
|
|
174
|
+
- 注册工具数:**36**
|
|
163
175
|
- 斜杠命令 hint:`start [override]|resume [override]|pause|abort|status|report|mode <auto|manual>|setup|save|template [global|project]|add <id> <desc>|add-proposition <id> <概述>|list-propositions|methods|index|plan|lock|project [list|new <name>|<name>]|decisions|agents|paper [lang=zh|en] [format=both|md|tex] [force]`
|
|
164
176
|
|
|
165
177
|
### config.prefix
|
|
@@ -211,9 +223,15 @@ is experiential reference.
|
|
|
211
223
|
- vibe_math_new_project / vibe_math_set_project / vibe_math_list_projects — per-project folders.
|
|
212
224
|
- vibe_math_list_decisions / vibe_math_decide {id, action: approve|reject|override, verdict?} — resolve manual decisions.
|
|
213
225
|
- vibe_math_list_agents / vibe_math_message_agent / vibe_math_interrupt_agent — inspect / steer / interrupt subagents.
|
|
214
|
-
- vibe_math_lean_run / vibe_math_lean_archive / vibe_math_lean_lib
|
|
215
|
-
verification (execute / archive / list the reuse
|
|
216
|
-
|
|
226
|
+
- vibe_math_lean_run / vibe_math_lean_archive / vibe_math_lean_lib / vibe_math_lean_read /
|
|
227
|
+
vibe_math_lean_job — Lean formal verification (execute / archive / list + read the reuse
|
|
228
|
+
library / inspect or wait for a background compile job). leanAsync=true (default) queues the
|
|
229
|
+
compile and returns a jobId; nothing counts as proved until that job settles ok (so "有把握"
|
|
230
|
+
is not the same as "已验证"). The scheduler's child agents use them too; they work in every
|
|
231
|
+
mode.
|
|
232
|
+
|
|
233
|
+
- math_computation {op: probe|run|receipt|install, engine, mode: code|file|expr, …} — 先 probe 预检引擎/包/许可,再调引擎计算并把脚本与输出归档成可复核回执;缺引擎/缺包只报告与给安装指引/计划;shell 兜底不算归档。
|
|
234
|
+
- 归档→编辑→重跑:mode:'code' 的脚本原件在回执的 scriptPath(Computation/<id>/script.<ext>),你可以用普通文件工具打开并编辑它;编辑后用 mode:'file' 指向它重跑,会写出一份**新回执/新 attempt**(含新的 scriptHash)。**旧回执对修改后的代码无效**——报告里必须引用与当前代码哈希一致的那份回执;工具会在 scriptChanged / scriptChangedDuringRun 为 true 时显式告警。
|
|
217
235
|
|
|
218
236
|
A /vibe slash command mirrors the main controls (/vibe start [override]|resume [override]|pause|abort|status|report|mode
|
|
219
237
|
<auto|manual>|setup|save|template [global|project]|add <id> <desc>|add-proposition <id> <概述>|list-propositions|methods|index|plan|lock|project [list|new <name>|<name>]|decisions|agents|paper [lang=zh|en] [format=both|md|tex] [force]).
|
|
@@ -241,7 +259,7 @@ LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
|
241
259
|
explicit, reasoned blocker record; the scheduler is never wedged by it).
|
|
242
260
|
- Paths: work file Formal/<id>.lean; archived proof Verified/Lean/<id>.lean; reusable
|
|
243
261
|
definitions VibeMath/Formal/Lib/; proved lemmas VibeMath/Formal/Proved/.
|
|
244
|
-
- The toolchain knobs leanCommand / leanArgs / leanTimeoutMs are tunable as well
|
|
262
|
+
- The toolchain knobs leanCommand / leanArgs / leanTimeoutMs / leanAsync / leanInitiative / leanSearchPaths / leanJobsMaxParallel are tunable as well
|
|
245
263
|
(e.g. leanCommand='lake' with leanArgs=['env','lean']); a missing Lean binary is
|
|
246
264
|
reported as LEAN_NOT_FOUND and still lets the code be written and archived.
|
|
247
265
|
- vibe_math_status / vibe_math_report show the mode, the per-object formal status and the
|
|
@@ -299,9 +317,15 @@ is experiential reference.
|
|
|
299
317
|
- vibe_math_new_project / vibe_math_set_project / vibe_math_list_projects — per-project folders.
|
|
300
318
|
- vibe_math_list_decisions / vibe_math_decide {id, action: approve|reject|override, verdict?} — resolve manual decisions.
|
|
301
319
|
- vibe_math_list_agents / vibe_math_message_agent / vibe_math_interrupt_agent — inspect / steer / interrupt subagents.
|
|
302
|
-
- vibe_math_lean_run / vibe_math_lean_archive / vibe_math_lean_lib
|
|
303
|
-
verification (execute / archive / list the reuse
|
|
304
|
-
|
|
320
|
+
- vibe_math_lean_run / vibe_math_lean_archive / vibe_math_lean_lib / vibe_math_lean_read /
|
|
321
|
+
vibe_math_lean_job — Lean formal verification (execute / archive / list + read the reuse
|
|
322
|
+
library / inspect or wait for a background compile job). leanAsync=true (default) queues the
|
|
323
|
+
compile and returns a jobId; nothing counts as proved until that job settles ok (so "有把握"
|
|
324
|
+
is not the same as "已验证"). The scheduler's child agents use them too; they work in every
|
|
325
|
+
mode.
|
|
326
|
+
|
|
327
|
+
- math_computation {op: probe|run|receipt|install, engine, mode: code|file|expr, …} — 先 probe 预检引擎/包/许可,再调引擎计算并把脚本与输出归档成可复核回执;缺引擎/缺包只报告与给安装指引/计划;shell 兜底不算归档。
|
|
328
|
+
- 归档→编辑→重跑:mode:'code' 的脚本原件在回执的 scriptPath(Computation/<id>/script.<ext>),你可以用普通文件工具打开并编辑它;编辑后用 mode:'file' 指向它重跑,会写出一份**新回执/新 attempt**(含新的 scriptHash)。**旧回执对修改后的代码无效**——报告里必须引用与当前代码哈希一致的那份回执;工具会在 scriptChanged / scriptChangedDuringRun 为 true 时显式告警。
|
|
305
329
|
|
|
306
330
|
A /vibe slash command mirrors the main controls (/vibe start [override]|resume [override]|pause|abort|status|report|mode
|
|
307
331
|
<auto|manual>|setup|save|template [global|project]|add <id> <desc>|add-proposition <id> <概述>|list-propositions|methods|index|plan|lock|project [list|new <name>|<name>]|decisions|agents|paper [lang=zh|en] [format=both|md|tex] [force]).
|
|
@@ -329,7 +353,7 @@ LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
|
329
353
|
explicit, reasoned blocker record; the scheduler is never wedged by it).
|
|
330
354
|
- Paths: work file Formal/<id>.lean; archived proof Verified/Lean/<id>.lean; reusable
|
|
331
355
|
definitions VibeMath/Formal/Lib/; proved lemmas VibeMath/Formal/Proved/.
|
|
332
|
-
- The toolchain knobs leanCommand / leanArgs / leanTimeoutMs are tunable as well
|
|
356
|
+
- The toolchain knobs leanCommand / leanArgs / leanTimeoutMs / leanAsync / leanInitiative / leanSearchPaths / leanJobsMaxParallel are tunable as well
|
|
333
357
|
(e.g. leanCommand='lake' with leanArgs=['env','lean']); a missing Lean binary is
|
|
334
358
|
reported as LEAN_NOT_FOUND and still lets the code be written and archived.
|
|
335
359
|
- vibe_math_status / vibe_math_report show the mode, the per-object formal status and the
|
|
@@ -341,7 +365,7 @@ When the user asks about progress, call vibe_math_report and summarize in plain
|
|
|
341
365
|
|
|
342
366
|
## vibe-math-v4
|
|
343
367
|
|
|
344
|
-
- 注册工具数:**
|
|
368
|
+
- 注册工具数:**34**
|
|
345
369
|
- 斜杠命令 hint:`configure|start|resume|pause|abort|status|report|message <to|all> <content>|meeting|paper [lang=zh|en] [format=md|tex|both] [editor=office|resident:<id>] [force]|members|add|remove|set`
|
|
346
370
|
|
|
347
371
|
### config.prefix
|
|
@@ -391,15 +415,20 @@ Data lives under {{cwd}}/VibeMath/Projects/<project>/:
|
|
|
391
415
|
Main controls (recommended flow: configure FIRST, then start):
|
|
392
416
|
- vibe_v4_configure {project?, problem?, params?} — create/configure the project (name, problem, params) WITHOUT starting a run; set everything here first.
|
|
393
417
|
- vibe_v4_start {problem?, residentCount?, seedDirections?} — begin the run (spawn residents, brainstorm). If problem was configured, omit it.
|
|
394
|
-
- vibe_v4_set {residentCount, compactThreshold, compactAfterRounds, meetingKeepEvery, maxParallel, activityTimeoutMs, stallAutoMeetingMs, verdictMaxRounds, provider, model, residentPersona, toolAllow, toolDeny, formalVerify, leanCommand, leanArgs, leanTimeoutMs, finalPaper, paperFormat, paperLanguage, paperCompilePdf, paperEditor, paperLatexCommand} — tune params (persisted to the settings file). provider/model override the residents' LLM route (empty = they inherit YOUR model/provider); toolAllow/toolDeny are per-resident tool permissions (empty = they inherit all tools). stallAutoMeetingMs is the stalled-group auto-sync-meeting threshold (分级保活 B). formalVerify (off|encourage|require, default off) enables Lean formal verification; leanCommand/leanArgs/leanTimeoutMs configure the toolchain. The FINAL PAPER is tuned by finalPaper (true|false, default true), paperFormat (both|md|tex), paperLanguage (zh|en), paperCompilePdf (true|false), paperEditor (office|resident:<id>, default office = the human/assistant side finalises) and paperLatexCommand.
|
|
418
|
+
- vibe_v4_set {residentCount, compactThreshold, compactAfterRounds, meetingKeepEvery, maxParallel, activityTimeoutMs, stallAutoMeetingMs, verdictMaxRounds, provider, model, residentPersona, toolAllow, toolDeny, formalVerify, leanCommand, leanArgs, leanTimeoutMs, leanAsync, leanInitiative, leanSearchPaths, leanJobsMaxParallel, finalPaper, paperFormat, paperLanguage, paperCompilePdf, paperEditor, paperLatexCommand} — tune params (persisted to the settings file). provider/model override the residents' LLM route (empty = they inherit YOUR model/provider); toolAllow/toolDeny are per-resident tool permissions (empty = they inherit all tools). stallAutoMeetingMs is the stalled-group auto-sync-meeting threshold (分级保活 B). formalVerify (off|encourage|require, default off) enables Lean formal verification; leanCommand/leanArgs/leanTimeoutMs configure the toolchain; leanAsync (true|false, default true) queues each compile as a background job (the tool returns {async:{jobId,state}} and the result lands later), leanInitiative (off|normal|eager, default normal) says how PROACTIVE the group is while working (independent of what formalVerify requires at voting time), leanSearchPaths adds extra compiler search roots (injected BEFORE the automatic VibeMath root), and leanJobsMaxParallel (default 1) caps concurrent compiles. The FINAL PAPER is tuned by finalPaper (true|false, default true), paperFormat (both|md|tex), paperLanguage (zh|en), paperCompilePdf (true|false), paperEditor (office|resident:<id>, default office = the human/assistant side finalises) and paperLatexCommand.
|
|
395
419
|
- vibe_v4_resume / vibe_v4_pause / vibe_v4_abort / vibe_v4_status / vibe_v4_report.
|
|
396
420
|
- vibe_v4_message {to|all, content} — inject a message to a resident (human/assistant intervention).
|
|
397
421
|
- vibe_v4_meeting {agenda} — force a meeting.
|
|
398
422
|
- vibe_v4_add_member {direction?} / vibe_v4_remove_member {id} — add / close a resident.
|
|
399
423
|
- vibe_v4_list_members — list residents.
|
|
400
|
-
- vibe_v4_lean_run / vibe_v4_lean_archive / vibe_v4_lean_lib
|
|
401
|
-
(execute / archive / list the reuse library
|
|
424
|
+
- vibe_v4_lean_run / vibe_v4_lean_archive / vibe_v4_lean_lib / vibe_v4_lean_read / vibe_v4_lean_job
|
|
425
|
+
— Lean formal verification (execute or ENQUEUE / archive / list the reuse library + background jobs /
|
|
426
|
+
read one archived library file verbatim / check or wait for a background job). Residents use them
|
|
427
|
+
too; they work in every mode.
|
|
402
428
|
- vibe_v4_formal_report — human-readable Lean formal-verification mirror (mode, Lean-passed
|
|
429
|
+
- math_computation {op: probe|run|receipt|install, engine, mode: code|file|expr, …} — 先 probe 预检引擎/包/许可,再调引擎计算并把脚本与输出归档成可复核回执;缺引擎/缺包只报告与给安装指引/计划;shell 兜底不算归档。
|
|
430
|
+
- 归档→编辑→重跑:mode:'code' 的脚本原件在回执的 scriptPath(Computation/<id>/script.<ext>),你可以用普通文件工具打开并编辑它;编辑后用 mode:'file' 指向它重跑,会写出一份**新回执/新 attempt**(含新的 scriptHash)。**旧回执对修改后的代码无效**——报告里必须引用与当前代码哈希一致的那份回执;工具会在 scriptChanged / scriptChangedDuringRun 为 true 时显式告警。
|
|
431
|
+
- Archive -> edit -> re-run: for mode:'code' the script original is at the receipt's scriptPath (Computation/<id>/script.<ext>) and you may open/edit it with your normal file tools; after editing, re-run it with mode:'file' to write a NEW receipt/attempt with a NEW scriptHash. **An old receipt is NOT evidence for edited code** - cite the receipt whose scriptHash matches the current code; the tool warns explicitly via scriptChanged / scriptChangedDuringRun.
|
|
403
432
|
objects, recorded blockers, formalization TODO, library paths).
|
|
404
433
|
- vibe_v4_prompts {which: brainstorm|normal|heartbeat|meeting|verify|coreRules, member?, target?, stage?} — read the exact prompt text a resident would receive (prompt auditing; prompt text is the product).
|
|
405
434
|
A /v4 slash command mirrors the main controls (configure|start [problem]|resume|pause|abort|status|report|message <to|all> <content>|meeting|paper [lang=zh|en] [format=md|tex|both] [editor=office|resident:<id>] [force]|members|add|remove|set <k=v>...).
|
|
@@ -433,6 +462,15 @@ LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
|
433
462
|
explicit, reasoned blocker record; the run is never wedged by it).
|
|
434
463
|
- Paths: work file Formal/<id>.lean; archived proof Verified/Lean/<id>.lean; reusable
|
|
435
464
|
definitions VibeMath/Formal/Lib/; proved lemmas VibeMath/Formal/Proved/.
|
|
465
|
+
- Reuse FIRST: call vibe_v4_lean_lib (and vibe_v4_lean_read for the verbatim text) BEFORE writing a new
|
|
466
|
+
definition; reuse an archived file with `import Formal.Lib.<name>` / `import Formal.Proved.<name>`
|
|
467
|
+
(the framework passes --search-path <VibeMath root>, so VibeMath is the module root).
|
|
468
|
+
- Choose WHAT to formalize with three criteria: (1) valuable or likely reusable, (2) rather key or
|
|
469
|
+
necessary, (3) you are confident in the statement — when you are NOT confident, record blocked
|
|
470
|
+
with the difficulty instead of hiding uncertainty behind a formalization.
|
|
471
|
+
- Compiles are BACKGROUND JOBS by default: a queued job is not a verification. An object only becomes
|
|
472
|
+
Lean-passed once its job settles green; the result is announced in the next prompt and visible via
|
|
473
|
+
vibe_v4_lean_lib.jobs / vibe_v4_lean_job (waitMs).
|
|
436
474
|
- The toolchain knobs leanCommand / leanArgs / leanTimeoutMs are tunable as well
|
|
437
475
|
(e.g. leanCommand='lake' with leanArgs=['env','lean']); a missing Lean binary is
|
|
438
476
|
reported as LEAN_NOT_FOUND and still lets the code be written and archived.
|
|
@@ -490,15 +528,20 @@ Data lives under {{cwd}}/VibeMath/Projects/<project>/:
|
|
|
490
528
|
Main controls (recommended flow: configure FIRST, then start):
|
|
491
529
|
- vibe_v4_configure {project?, problem?, params?} — create/configure the project (name, problem, params) WITHOUT starting a run; set everything here first.
|
|
492
530
|
- vibe_v4_start {problem?, residentCount?, seedDirections?} — begin the run (spawn residents, brainstorm). If problem was configured, omit it.
|
|
493
|
-
- vibe_v4_set {residentCount, compactThreshold, compactAfterRounds, meetingKeepEvery, maxParallel, activityTimeoutMs, stallAutoMeetingMs, verdictMaxRounds, provider, model, residentPersona, toolAllow, toolDeny, formalVerify, leanCommand, leanArgs, leanTimeoutMs, finalPaper, paperFormat, paperLanguage, paperCompilePdf, paperEditor, paperLatexCommand} — tune params (persisted to the settings file). provider/model override the residents' LLM route (empty = they inherit YOUR model/provider); toolAllow/toolDeny are per-resident tool permissions (empty = they inherit all tools). stallAutoMeetingMs is the stalled-group auto-sync-meeting threshold (分级保活 B). formalVerify (off|encourage|require, default off) enables Lean formal verification; leanCommand/leanArgs/leanTimeoutMs configure the toolchain. The FINAL PAPER is tuned by finalPaper (true|false, default true), paperFormat (both|md|tex), paperLanguage (zh|en), paperCompilePdf (true|false), paperEditor (office|resident:<id>, default office = the human/assistant side finalises) and paperLatexCommand.
|
|
531
|
+
- vibe_v4_set {residentCount, compactThreshold, compactAfterRounds, meetingKeepEvery, maxParallel, activityTimeoutMs, stallAutoMeetingMs, verdictMaxRounds, provider, model, residentPersona, toolAllow, toolDeny, formalVerify, leanCommand, leanArgs, leanTimeoutMs, leanAsync, leanInitiative, leanSearchPaths, leanJobsMaxParallel, finalPaper, paperFormat, paperLanguage, paperCompilePdf, paperEditor, paperLatexCommand} — tune params (persisted to the settings file). provider/model override the residents' LLM route (empty = they inherit YOUR model/provider); toolAllow/toolDeny are per-resident tool permissions (empty = they inherit all tools). stallAutoMeetingMs is the stalled-group auto-sync-meeting threshold (分级保活 B). formalVerify (off|encourage|require, default off) enables Lean formal verification; leanCommand/leanArgs/leanTimeoutMs configure the toolchain; leanAsync (true|false, default true) queues each compile as a background job (the tool returns {async:{jobId,state}} and the result lands later), leanInitiative (off|normal|eager, default normal) says how PROACTIVE the group is while working (independent of what formalVerify requires at voting time), leanSearchPaths adds extra compiler search roots (injected BEFORE the automatic VibeMath root), and leanJobsMaxParallel (default 1) caps concurrent compiles. The FINAL PAPER is tuned by finalPaper (true|false, default true), paperFormat (both|md|tex), paperLanguage (zh|en), paperCompilePdf (true|false), paperEditor (office|resident:<id>, default office = the human/assistant side finalises) and paperLatexCommand.
|
|
494
532
|
- vibe_v4_resume / vibe_v4_pause / vibe_v4_abort / vibe_v4_status / vibe_v4_report.
|
|
495
533
|
- vibe_v4_message {to|all, content} — inject a message to a resident (human/assistant intervention).
|
|
496
534
|
- vibe_v4_meeting {agenda} — force a meeting.
|
|
497
535
|
- vibe_v4_add_member {direction?} / vibe_v4_remove_member {id} — add / close a resident.
|
|
498
536
|
- vibe_v4_list_members — list residents.
|
|
499
|
-
- vibe_v4_lean_run / vibe_v4_lean_archive / vibe_v4_lean_lib
|
|
500
|
-
(execute / archive / list the reuse library
|
|
537
|
+
- vibe_v4_lean_run / vibe_v4_lean_archive / vibe_v4_lean_lib / vibe_v4_lean_read / vibe_v4_lean_job
|
|
538
|
+
— Lean formal verification (execute or ENQUEUE / archive / list the reuse library + background jobs /
|
|
539
|
+
read one archived library file verbatim / check or wait for a background job). Residents use them
|
|
540
|
+
too; they work in every mode.
|
|
501
541
|
- vibe_v4_formal_report — human-readable Lean formal-verification mirror (mode, Lean-passed
|
|
542
|
+
- math_computation {op: probe|run|receipt|install, engine, mode: code|file|expr, …} — 先 probe 预检引擎/包/许可,再调引擎计算并把脚本与输出归档成可复核回执;缺引擎/缺包只报告与给安装指引/计划;shell 兜底不算归档。
|
|
543
|
+
- 归档→编辑→重跑:mode:'code' 的脚本原件在回执的 scriptPath(Computation/<id>/script.<ext>),你可以用普通文件工具打开并编辑它;编辑后用 mode:'file' 指向它重跑,会写出一份**新回执/新 attempt**(含新的 scriptHash)。**旧回执对修改后的代码无效**——报告里必须引用与当前代码哈希一致的那份回执;工具会在 scriptChanged / scriptChangedDuringRun 为 true 时显式告警。
|
|
544
|
+
- Archive -> edit -> re-run: for mode:'code' the script original is at the receipt's scriptPath (Computation/<id>/script.<ext>) and you may open/edit it with your normal file tools; after editing, re-run it with mode:'file' to write a NEW receipt/attempt with a NEW scriptHash. **An old receipt is NOT evidence for edited code** - cite the receipt whose scriptHash matches the current code; the tool warns explicitly via scriptChanged / scriptChangedDuringRun.
|
|
502
545
|
objects, recorded blockers, formalization TODO, library paths).
|
|
503
546
|
- vibe_v4_prompts {which: brainstorm|normal|heartbeat|meeting|verify|coreRules, member?, target?, stage?} — read the exact prompt text a resident would receive (prompt auditing; prompt text is the product).
|
|
504
547
|
A /v4 slash command mirrors the main controls (configure|start [problem]|resume|pause|abort|status|report|message <to|all> <content>|meeting|paper [lang=zh|en] [format=md|tex|both] [editor=office|resident:<id>] [force]|members|add|remove|set <k=v>...).
|
|
@@ -532,6 +575,15 @@ LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
|
532
575
|
explicit, reasoned blocker record; the run is never wedged by it).
|
|
533
576
|
- Paths: work file Formal/<id>.lean; archived proof Verified/Lean/<id>.lean; reusable
|
|
534
577
|
definitions VibeMath/Formal/Lib/; proved lemmas VibeMath/Formal/Proved/.
|
|
578
|
+
- Reuse FIRST: call vibe_v4_lean_lib (and vibe_v4_lean_read for the verbatim text) BEFORE writing a new
|
|
579
|
+
definition; reuse an archived file with `import Formal.Lib.<name>` / `import Formal.Proved.<name>`
|
|
580
|
+
(the framework passes --search-path <VibeMath root>, so VibeMath is the module root).
|
|
581
|
+
- Choose WHAT to formalize with three criteria: (1) valuable or likely reusable, (2) rather key or
|
|
582
|
+
necessary, (3) you are confident in the statement — when you are NOT confident, record blocked
|
|
583
|
+
with the difficulty instead of hiding uncertainty behind a formalization.
|
|
584
|
+
- Compiles are BACKGROUND JOBS by default: a queued job is not a verification. An object only becomes
|
|
585
|
+
Lean-passed once its job settles green; the result is announced in the next prompt and visible via
|
|
586
|
+
vibe_v4_lean_lib.jobs / vibe_v4_lean_job (waitMs).
|
|
535
587
|
- The toolchain knobs leanCommand / leanArgs / leanTimeoutMs are tunable as well
|
|
536
588
|
(e.g. leanCommand='lake' with leanArgs=['env','lean']); a missing Lean binary is
|
|
537
589
|
reported as LEAN_NOT_FOUND and still lets the code be written and archived.
|
|
@@ -545,7 +597,7 @@ When the user asks about progress, call vibe_v4_report and summarize in plain la
|
|
|
545
597
|
|
|
546
598
|
## vibe-math-v5
|
|
547
599
|
|
|
548
|
-
- 注册工具数:**
|
|
600
|
+
- 注册工具数:**39**
|
|
549
601
|
- 斜杠命令 hint:`configure|start|resume|pause|stop|status|report|members|message|meeting|hire|fire|add|remove|set|paper [lang=en] [format=tex] [editor=office] [force]`
|
|
550
602
|
|
|
551
603
|
### config.prefix
|
|
@@ -620,9 +672,12 @@ Main controls (recommended flow: configure FIRST, then start):
|
|
|
620
672
|
- vibe_v5_add_researcher / vibe_v5_remove_researcher — OFFICE only: add or dismiss a PERMANENT researcher (the academician can only propose those).
|
|
621
673
|
- vibe_v5_paper {lang?, format?, editor?, force?} — write the FINAL PAPER: the permanent staff write their own parts, cross-review each other, then the editor finalises. lang/format/editor override this one paper only.
|
|
622
674
|
- vibe_v5_finalize_paper {decision, note, conclusion?} — OFFICE only when paperEditor=office: consult the institute first (>=1 office message + >=1 meeting), then finalise and put the conclusion in note.
|
|
623
|
-
- vibe_v5_lean_run / vibe_v5_lean_archive / vibe_v5_lean_lib — Lean formal
|
|
624
|
-
verification (execute / archive / list the reuse library
|
|
625
|
-
too; they work in every mode.
|
|
675
|
+
- vibe_v5_lean_run / vibe_v5_lean_archive / vibe_v5_lean_lib / vibe_v5_lean_read / vibe_v5_lean_job — Lean formal
|
|
676
|
+
verification (execute / archive / list the reuse library / read an archived file
|
|
677
|
+
verbatim / watch the background compile queue). Members use them too; they work in every mode.
|
|
678
|
+
- math_computation {op: probe|run|receipt|install, engine, mode: code|file|expr, …} — 先 probe 预检引擎/包/许可,再调引擎计算并把脚本与输出归档成可复核回执;缺引擎/缺包只报告与给安装指引/计划;shell 兜底不算归档。
|
|
679
|
+
- 归档→编辑→重跑:mode:'code' 的脚本原件在回执的 scriptPath(Computation/<id>/script.<ext>),你可以用普通文件工具打开并编辑它;编辑后用 mode:'file' 指向它重跑,会写出一份**新回执/新 attempt**(含新的 scriptHash)。**旧回执对修改后的代码无效**——报告里必须引用与当前代码哈希一致的那份回执;工具会在 scriptChanged / scriptChangedDuringRun 为 true 时显式告警。
|
|
680
|
+
- Archive -> edit -> re-run: for mode:'code' the script original is at the receipt's scriptPath (Computation/<id>/script.<ext>) and you may open/edit it with your normal file tools; after editing, re-run it with mode:'file' to write a NEW receipt/attempt with a NEW scriptHash. **An old receipt is NOT evidence for edited code** - cite the receipt whose scriptHash matches the current code; the tool warns explicitly via scriptChanged / scriptChangedDuringRun.
|
|
626
681
|
A /v5 slash command mirrors these (configure|start|resume|pause|stop|status|report|members|message|meeting|hire|fire|add|remove|set|paper).
|
|
627
682
|
|
|
628
683
|
LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
@@ -638,6 +693,19 @@ LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
|
638
693
|
reported as LEAN_NOT_FOUND, a host with no subprocess service as NO_SUBPROCESS
|
|
639
694
|
(on a timeout the process is terminated first), and in every one of those cases the
|
|
640
695
|
code can still be written down and archived.
|
|
696
|
+
- Daily eagerness is a SEPARATE knob: leanInitiative 'off' (no daily drive) |
|
|
697
|
+
'normal' (default; the reminder rides with formalVerify) | 'eager' (push valuable
|
|
698
|
+
small lemmas/propositions/definitions into the library even in ordinary rounds).
|
|
699
|
+
- Reuse first: check vibe_v5_lean_lib BEFORE writing a definition, reuse archived
|
|
700
|
+
content via `import Formal.Lib.<name>` / `import Formal.Proved.<name>` (the compile
|
|
701
|
+
gets `--search-path <VibeMath root>`, preceded by any leanSearchPaths), or copy the
|
|
702
|
+
exact text with vibe_v5_lean_read. Re-archiving identical content is de-duplicated.
|
|
703
|
+
- Compiles are ASYNC by default (leanAsync=true, leanJobsMaxParallel default 1):
|
|
704
|
+
vibe_v5_lean_run / vibe_v5_lean_archive enqueue and return at once; watch them with
|
|
705
|
+
vibe_v5_lean_job {jobId, waitMs} or vibe_v5_lean_lib.jobs, and NEVER treat an object
|
|
706
|
+
as passed before its job settles ok (exit 0, same content, same build context).
|
|
707
|
+
- Only formalize what you are confident about; when unsure record kind='blocked' with a
|
|
708
|
+
reason instead of dressing uncertainty up as a formalization.
|
|
641
709
|
- vibe_v5_status / vibe_v5_report show the mode, per-object formal status and the
|
|
642
710
|
formalization TODO. The framework never installs Lean and never judges fidelity.
|
|
643
711
|
|
|
@@ -721,9 +789,12 @@ Main controls (recommended flow: configure FIRST, then start):
|
|
|
721
789
|
- vibe_v5_add_researcher / vibe_v5_remove_researcher — OFFICE only: add or dismiss a PERMANENT researcher (the academician can only propose those).
|
|
722
790
|
- vibe_v5_paper {lang?, format?, editor?, force?} — write the FINAL PAPER: the permanent staff write their own parts, cross-review each other, then the editor finalises. lang/format/editor override this one paper only.
|
|
723
791
|
- vibe_v5_finalize_paper {decision, note, conclusion?} — OFFICE only when paperEditor=office: consult the institute first (>=1 office message + >=1 meeting), then finalise and put the conclusion in note.
|
|
724
|
-
- vibe_v5_lean_run / vibe_v5_lean_archive / vibe_v5_lean_lib — Lean formal
|
|
725
|
-
verification (execute / archive / list the reuse library
|
|
726
|
-
too; they work in every mode.
|
|
792
|
+
- vibe_v5_lean_run / vibe_v5_lean_archive / vibe_v5_lean_lib / vibe_v5_lean_read / vibe_v5_lean_job — Lean formal
|
|
793
|
+
verification (execute / archive / list the reuse library / read an archived file
|
|
794
|
+
verbatim / watch the background compile queue). Members use them too; they work in every mode.
|
|
795
|
+
- math_computation {op: probe|run|receipt|install, engine, mode: code|file|expr, …} — 先 probe 预检引擎/包/许可,再调引擎计算并把脚本与输出归档成可复核回执;缺引擎/缺包只报告与给安装指引/计划;shell 兜底不算归档。
|
|
796
|
+
- 归档→编辑→重跑:mode:'code' 的脚本原件在回执的 scriptPath(Computation/<id>/script.<ext>),你可以用普通文件工具打开并编辑它;编辑后用 mode:'file' 指向它重跑,会写出一份**新回执/新 attempt**(含新的 scriptHash)。**旧回执对修改后的代码无效**——报告里必须引用与当前代码哈希一致的那份回执;工具会在 scriptChanged / scriptChangedDuringRun 为 true 时显式告警。
|
|
797
|
+
- Archive -> edit -> re-run: for mode:'code' the script original is at the receipt's scriptPath (Computation/<id>/script.<ext>) and you may open/edit it with your normal file tools; after editing, re-run it with mode:'file' to write a NEW receipt/attempt with a NEW scriptHash. **An old receipt is NOT evidence for edited code** - cite the receipt whose scriptHash matches the current code; the tool warns explicitly via scriptChanged / scriptChangedDuringRun.
|
|
727
798
|
A /v5 slash command mirrors these (configure|start|resume|pause|stop|status|report|members|message|meeting|hire|fire|add|remove|set|paper).
|
|
728
799
|
|
|
729
800
|
LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
@@ -739,6 +810,19 @@ LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
|
|
|
739
810
|
reported as LEAN_NOT_FOUND, a host with no subprocess service as NO_SUBPROCESS
|
|
740
811
|
(on a timeout the process is terminated first), and in every one of those cases the
|
|
741
812
|
code can still be written down and archived.
|
|
813
|
+
- Daily eagerness is a SEPARATE knob: leanInitiative 'off' (no daily drive) |
|
|
814
|
+
'normal' (default; the reminder rides with formalVerify) | 'eager' (push valuable
|
|
815
|
+
small lemmas/propositions/definitions into the library even in ordinary rounds).
|
|
816
|
+
- Reuse first: check vibe_v5_lean_lib BEFORE writing a definition, reuse archived
|
|
817
|
+
content via `import Formal.Lib.<name>` / `import Formal.Proved.<name>` (the compile
|
|
818
|
+
gets `--search-path <VibeMath root>`, preceded by any leanSearchPaths), or copy the
|
|
819
|
+
exact text with vibe_v5_lean_read. Re-archiving identical content is de-duplicated.
|
|
820
|
+
- Compiles are ASYNC by default (leanAsync=true, leanJobsMaxParallel default 1):
|
|
821
|
+
vibe_v5_lean_run / vibe_v5_lean_archive enqueue and return at once; watch them with
|
|
822
|
+
vibe_v5_lean_job {jobId, waitMs} or vibe_v5_lean_lib.jobs, and NEVER treat an object
|
|
823
|
+
as passed before its job settles ok (exit 0, same content, same build context).
|
|
824
|
+
- Only formalize what you are confident about; when unsure record kind='blocked' with a
|
|
825
|
+
reason instead of dressing uncertainty up as a formalization.
|
|
742
826
|
- vibe_v5_status / vibe_v5_report show the mode, per-object formal status and the
|
|
743
827
|
formalization TODO. The framework never installs Lean and never judges fidelity.
|
|
744
828
|
|