dsh-github-copilot 0.4.0-alpha.99 → 0.4.0

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.
Files changed (109) hide show
  1. package/AGENTS.md +96 -16
  2. package/CHANGELOG.md +211 -0
  3. package/CONTRIBUTING.md +9 -0
  4. package/README.md +106 -18
  5. package/README.zh.md +97 -18
  6. package/deployment-baseline.json +227 -8
  7. package/docs/account-management-experience.md +143 -0
  8. package/docs/auto-high-cost.md +154 -0
  9. package/docs/auto-iteration-review.json +22 -0
  10. package/docs/auto-review-sampling.md +60 -0
  11. package/docs/auto-task-routing.md +8 -6
  12. package/docs/automatic-model-routing.md +33 -5
  13. package/docs/copilot-accounts.md +376 -0
  14. package/docs/copilot-compaction.md +23 -2
  15. package/docs/copilot-usage.md +24 -4
  16. package/docs/evidence-driven-iteration.md +86 -0
  17. package/docs/images/copilot-accounts-credits.png +0 -0
  18. package/docs/images/copilot-accounts.png +0 -0
  19. package/docs/images/copilot-current-provenance.json +39 -0
  20. package/docs/images/copilot-model-preferences.png +0 -0
  21. package/docs/images/copilot-search-routing.png +0 -0
  22. package/docs/images/copilot-session-continuation.png +0 -0
  23. package/docs/manual-compaction-recovery.md +50 -7
  24. package/docs/mockups/copilot-account-dropdown.html +71 -0
  25. package/docs/mockups/copilot-account-management.html +393 -0
  26. package/docs/model-compatibility-acceptance.md +34 -16
  27. package/docs/official-first-020-rc2.md +1 -0
  28. package/docs/plugin-diagnostics.md +452 -0
  29. package/docs/session-continuation.md +141 -0
  30. package/docs/session-search-routing.md +1 -1
  31. package/docs/single-route-migration.md +5 -4
  32. package/lib/client.js +4415 -846
  33. package/lib/client.js.map +1 -1
  34. package/lib/diagnostics-analysis-io.js +78 -0
  35. package/lib/diagnostics-analysis.js +425 -0
  36. package/lib/diagnostics-types-FVf6zjob.js +182 -0
  37. package/lib/index.js +10356 -6020
  38. package/lib/manual-compaction-recovery.js +163 -38
  39. package/lib/remote.js +593 -34
  40. package/lib/routed-web.js +1 -1
  41. package/lib/{search-routing-jzNYveyK.js → search-routing-CV2hs8as.js} +139 -139
  42. package/lib/types/account-dropdown.d.ts +22 -0
  43. package/lib/types/account-model-auth.d.ts +1 -1
  44. package/lib/types/authorization-code-clipboard.d.ts +4 -0
  45. package/lib/types/authorization-controller.d.ts +2 -0
  46. package/lib/types/auto-allocation-card.d.ts +9 -0
  47. package/lib/types/auto-allocation-evidence.d.ts +38 -0
  48. package/lib/types/auto-allocation.d.ts +24 -0
  49. package/lib/types/auto-model-host.d.ts +2 -0
  50. package/lib/types/auto-model-routing.d.ts +6 -2
  51. package/lib/types/auto-review-sampling.d.ts +6 -0
  52. package/lib/types/auto-task-classifier.d.ts +1 -1
  53. package/lib/types/background-compaction.d.ts +2 -2
  54. package/lib/types/client.d.ts +19 -3
  55. package/lib/types/compaction-continuation-ui.d.ts +29 -0
  56. package/lib/types/compaction-continuation.d.ts +48 -0
  57. package/lib/types/compaction-input-estimate.d.ts +24 -0
  58. package/lib/types/compaction-replay.d.ts +18 -0
  59. package/lib/types/composer-notice-style.d.ts +7 -0
  60. package/lib/types/config.d.ts +42 -2
  61. package/lib/types/context-evidence-ui.d.ts +18 -1
  62. package/lib/types/context-evidence.d.ts +152 -4
  63. package/lib/types/copilot-account-presentation.d.ts +7 -0
  64. package/lib/types/copilot-accounts-card.d.ts +43 -0
  65. package/lib/types/copilot-accounts-host.d.ts +93 -0
  66. package/lib/types/copilot-accounts-remote.d.ts +85 -0
  67. package/lib/types/copilot-accounts-types.d.ts +43 -0
  68. package/lib/types/copilot-auth.d.ts +3 -1
  69. package/lib/types/copilot-usage-card.d.ts +23 -0
  70. package/lib/types/copilot-usage-host.d.ts +9 -1
  71. package/lib/types/copilot-usage-remote.d.ts +2 -1
  72. package/lib/types/copilot-usage-types.d.ts +1 -0
  73. package/lib/types/diagnostics-analysis-io.d.ts +23 -0
  74. package/lib/types/diagnostics-analysis.d.ts +295 -0
  75. package/lib/types/diagnostics-card.d.ts +11 -0
  76. package/lib/types/diagnostics-client.d.ts +13 -0
  77. package/lib/types/diagnostics-collector.d.ts +36 -0
  78. package/lib/types/diagnostics-host.d.ts +72 -0
  79. package/lib/types/diagnostics-remote.d.ts +109 -0
  80. package/lib/types/diagnostics-types.d.ts +431 -0
  81. package/lib/types/manual-compaction-recovery.d.ts +18 -3
  82. package/lib/types/model-preferences-card.d.ts +1 -1
  83. package/lib/types/native-settings-style.d.ts +10 -0
  84. package/lib/types/parent-model-follow-card.d.ts +2 -11
  85. package/lib/types/preview-provider.d.ts +4 -0
  86. package/lib/types/preview-route.d.ts +7 -5
  87. package/lib/types/remote.d.ts +9 -6
  88. package/lib/types/replay-recovery-host.d.ts +2 -2
  89. package/lib/types/replay-recovery-ui.d.ts +3 -1
  90. package/lib/types/request-body-evidence.d.ts +3 -1
  91. package/lib/types/responses-replay-compat.d.ts +2 -1
  92. package/lib/types/responses-request-compression.d.ts +33 -0
  93. package/lib/types/session-accounts-controller.d.ts +20 -0
  94. package/lib/types/session-accounts-host.d.ts +33 -0
  95. package/lib/types/session-accounts-remote.d.ts +114 -0
  96. package/lib/types/session-accounts.d.ts +38 -0
  97. package/lib/types/session-continuation-host.d.ts +45 -0
  98. package/lib/types/session-continuation-remote.d.ts +22 -0
  99. package/lib/types/session-continuation-types.d.ts +62 -0
  100. package/lib/types/session-continuation-ui.d.ts +49 -0
  101. package/lib/types/session-continuation.d.ts +14 -0
  102. package/lib/types/turn-request-models.d.ts +55 -0
  103. package/lib/types/turn-selection-card.d.ts +15 -1
  104. package/lib/types/turn-selection-host.d.ts +23 -0
  105. package/lib/types/turn-selection-remote.d.ts +4 -0
  106. package/lib/types/turn-selection.d.ts +19 -4
  107. package/package.json +19 -3
  108. package/scripts/analyze-diagnostics.mjs +91 -0
  109. package/scripts/iteration-review.mjs +53 -0
@@ -0,0 +1,452 @@
1
+ # Local diagnostics: account and compaction pilot
2
+
3
+ Roadmap: [#347](https://github.com/cloga/dsh-github-copilot/issues/347).
4
+ Phase 0: [#354](https://github.com/cloga/dsh-github-copilot/issues/354).
5
+ Phases 1–2: [#356](https://github.com/cloga/dsh-github-copilot/issues/356).
6
+ Reviewed source baseline: plugin alpha.123 and official DSH
7
+ `0.2.0-rc.2`, commit `639ed015397290b3745d163aafe02ffee4aa3f84`.
8
+
9
+ **Runtime pilot, default off. Publication, installation, loaded Host and enabled
10
+ collection are separate states.** This feature does not install itself, restart
11
+ Desktop, enable a live profile, create a daily task or upload observations.
12
+ No historical failure rate is reconstructed.
13
+
14
+ ## Enable, read, pause and clear
15
+
16
+ After installing and loading the published build, open the plugin's detail
17
+ settings page and find **Local diagnostics**. On hosts without that page the
18
+ same additive control accompanies the existing Models/footer or settings-section
19
+ search-routing fallback. **Read status** must show ready storage; choose
20
+ **Enable local collection** to persist only `github-copilot.diagnosticsEnabled`
21
+ through native path-level Settings CAS and confirm readback. This enables
22
+ collection **and bounded local aggregate persistence**, not exports or uploads.
23
+ The normal installation does not change this default-off setting.
24
+
25
+ Use the plugin normally: account selection, Identity/Checking and compaction
26
+ will contribute independent Client/Host populations. **Read status** returns a
27
+ strict aggregate snapshot; **Review aggregate-only JSON → Prepare current JSON**
28
+ prepares a read-only copy for explicit review/sharing, never a download or upload.
29
+ The coding session does not automatically gain access to the running Host.
30
+ Pause preserves retained evidence and closes active observations as interrupted.
31
+ It remains available for a configured-on collector when storage is unavailable;
32
+ enabling still requires ready storage.
33
+ Clear requires a separate confirmation, advances an epoch and removes this
34
+ domain's aggregates without changing the enabled setting, accounts or history.
35
+ An uncertain write is not automatically replayed by the UI.
36
+
37
+ ## Storage and limits
38
+
39
+ The Host uses only the public storage-domain `global.get/set/close` seam and the
40
+ configured backend. The plugin pins Zod to the qualified published storage
41
+ type contract; no shared schemas, prototypes or Core artifacts are adapted.
42
+ The official JSON backend root is **DSH-home shared**, so
43
+ the plugin addresses a separate domain using the public `profileContext.name`,
44
+ encoded as a bounded storage-unit suffix. The technical profile name is used
45
+ only for addressing, never included in rows, Remotes or exported JSON. Names
46
+ must be 1–48 ASCII letters, digits, underscores or hyphens; missing/unsupported
47
+ profile context fails explicitly rather than sharing another profile's data.
48
+ One writer per profile is required: the public JSON backend does not establish
49
+ cross-process locking for simultaneous Hosts writing the same domain.
50
+
51
+ Retain UTC-hour rows for 14 days, at most 4,096 rows, 128 live Host handles,
52
+ 128 live Client handles and a 128-row Client report buffer. Strict fixed ASCII
53
+ dimensions make the encoded snapshot ceiling an exact **2 MiB UTF-8** bound.
54
+ Evict oldest-hour rows first and expose eviction, drops, saturation, Client
55
+ unconfirmed acknowledgements and clock rollback. Rollback does not silently
56
+ erase already observed future-hour rows. Counts across a window boundary or
57
+ an eviction cannot establish a precise failure rate.
58
+
59
+ The Host coalesces writes on a ten-second cadence; the Client reports safe
60
+ rows/pending-age samples and reads collection status on a ten-second cadence.
61
+ These lifecycle-bound timers do **not** perform analysis, query credentials,
62
+ change request deadlines or trigger repair. Abrupt process loss can lose the
63
+ unflushed tail. Saved unfinished Host groups become interrupted on reopen,
64
+ attributed to their starting build, not resumable work. Client pending samples
65
+ are historical periodic samples, **not** an outstanding-operation gauge.
66
+ Clear/pause/enable epoch fences prevent late accepted reports or handles from
67
+ restoring pre-control evidence.
68
+ Epoch exhaustion blocks enabling explicitly; pause/close still stop collection
69
+ and record saturation even when the fence cannot advance.
70
+
71
+ `dirty` and `persistedAt` distinguish in-memory from confirmed persistence.
72
+ Missing backend/profile, corruption and failed writes remain named unavailable
73
+ or error evidence; account/compaction work is not rejected or retried for a
74
+ diagnostics failure. No ad-hoc file, Settings log or fallback backend is used.
75
+ Corrupt media is neither backed up nor silently reset; Clear cannot repair a
76
+ domain that cannot open. Public domain open loads external media before schema
77
+ validation: plugin-owned snapshots are bounded, but this is **not** a disk quota
78
+ or a pre-read bound on externally tampered files.
79
+
80
+ ## Coverage and interpretation
81
+
82
+ Host coverage includes global switch, Session select/inherit, logical identity
83
+ reads and shared physical identity flights, native compaction brackets and
84
+ managed physical summary calls inside the separately selected recovery engine.
85
+ Cache, cooldown and joined-flight stages remain distinct. A global-switch
86
+ preflight identity validation is a stage, not another shared-flight denominator.
87
+ Only the wrapped Client account/identity operations are reported, independently
88
+ of Host observations; no IDs correlate the layers and no end-to-end latency or
89
+ paint completion is inferred. Component teardown settles its own pending
90
+ Checking observations, without cancelling the actual business request.
91
+
92
+ Outcomes distinguish success, cancellation with actual signal evidence, policy
93
+ rejection, revocation, environment fault, failure, interruption and unknown.
94
+ Concurrent validation stages record cumulative operation age, not exclusive
95
+ phase duration. A fulfilled error-shaped account view is not success.
96
+ Compaction success requires a matching observed compact-checkpoint **and**
97
+ successful end; summary fulfillment alone is not recovery success. No-checkpoint
98
+ ends are unknown, not guessed no-op/prune-only success. Commit evidence remains
99
+ separate if the end later fails. Selected managed-summary cancellation supplies
100
+ signal evidence; arbitrary native error text cannot establish cancellation.
101
+
102
+ Stock/custom-engine physical summaries, recovery opt-out/native-only summary
103
+ branches, OAuth Add/reauthorization/removal, quota, ordinary model refresh,
104
+ managed chat/search/classifier transport and unwrapped RPCs remain uncovered.
105
+ Native checkpoint observation is not filesystem durability evidence. Reporting
106
+ loss, crashes, disabled intervals and paging/coverage gaps must never be treated
107
+ as zero failures. No automatic daily analysis, issue creation, account/model
108
+ switch, policy change, retry, compaction, restart or upload is introduced.
109
+
110
+ The tests exercise bounded codecs, once-only settlement, control epochs,
111
+ component teardown, actual Client/Host gateway binding, unchanged public JSON
112
+ storage reopen/isolation/write failure/corruption and native SettingsForms
113
+ hidden-leaf persistence. Compaction observer envelopes are synthetic; they are
114
+ not production reliability measurements or proof of live recovery. Complete
115
+ release CI and installed/loaded/collection evidence remain separate.
116
+ The published Client gateway fixture captures the unchanged registration-bundle
117
+ factory in its own isolated realm, sharing the actual public Cordis identity.
118
+ It does not substitute tagged source for published bytes or attach to a live
119
+ Client module registry; factory qualification is not Desktop activation proof.
120
+
121
+ ## Explicit offline analysis (#368)
122
+
123
+ The package provides a separate, read-only analyzer for one explicitly selected
124
+ aggregate file. Build the package first, then provide an absolute input path, an
125
+ input mode, and a new local output path. The output is created exclusively and
126
+ is never printed to stdout or sent over the network:
127
+
128
+ ```sh
129
+ node scripts/analyze-diagnostics.mjs --input "ABSOLUTE_UNIT_FILE.json" --mode persisted-unit --profile-name "PROFILE_NAME" --output "NEW_LOCAL_REPORT.json"
130
+ ```
131
+
132
+ Replace the quoted placeholders before running; the input must be an absolute
133
+ path whose basename matches the profile-qualified unit filename.
134
+
135
+ For an explicitly reviewed Client/Host Remote view export, use
136
+ `--mode reviewed-view` and omit `--profile-name`. The persisted-unit mode
137
+ requires the unit basename and unit envelope to match the supplied 1–48
138
+ character ASCII DSH profile name; the name is mapped using the same public
139
+ profile-qualified storage-domain convention as the collector. The operator
140
+ must identify the file and profile. The tool never searches DSH homes,
141
+ enumerates storage units, opens a storage-domain writer or RPC, reads Settings,
142
+ credentials or history, or falls back between modes.
143
+
144
+ Both modes reuse the collector's strict Zod snapshot/view vocabulary. The
145
+ reader rejects symlinks and non-files, caps input at 8 MiB, opens only for
146
+ reading, decodes fatal UTF-8, and checks file identity, size and modification
147
+ metadata before and after reading. These checks bound and detect ordinary
148
+ concurrent changes; they do not attest the active Host's backend and cannot
149
+ eliminate every filesystem race. Errors expose fixed diagnostic codes, not
150
+ paths, rejected values or raw filesystem details. The strict versioned report
151
+ contains only validated dimensions, bounded counts and explicit gaps. An
152
+ existing output is never overwritten.
153
+
154
+ Populations stay separate by build version, Client/Host layer and operation,
155
+ including logical/physical pairs. Starts and terminal counters are reported
156
+ independently because UTC-hour aggregation and retention can censor either
157
+ side; pending-start subtraction and exact operation failure rates are never
158
+ calculated. Only terminal observations enter duration bins. Stage-age bins are
159
+ cumulative operation ages, not exclusive phase timings. Client pending samples
160
+ are repeated historical observations, not an outstanding-work gauge; persisted
161
+ Host pending rows describe the stored snapshot, not current live work.
162
+
163
+ Outcomes remain separate, including policy rejection, revocation,
164
+ environment-fault, unknown and interrupted. Safe sums that exceed JavaScript's
165
+ exact integer range become `null` with an overflow gap. Fixed semantic candidate
166
+ fingerprints contain only coverage version, layer, operation, metric and
167
+ allowlisted reason—never identity hashes. Candidates do not prove a regression,
168
+ persistent fault, unique impact or root cause. Reports do not claim calibrated
169
+ thresholds, trends, percentiles, effectiveness or live collection status from
170
+ a persisted-only file. Empty, stale, paused, dirty, saturated, dropped,
171
+ evicted, interrupted and clock-discontinuous evidence remains explicit;
172
+ missing account-switch or compaction observations are gaps, not zero failures.
173
+ A physical summary-call success is explicitly not a native checkpoint commit;
174
+ only the logical compaction population can label a matching checkpoint and end
175
+ as observed success.
176
+
177
+ The first locally qualified persisted snapshot produced a private descriptive
178
+ report outside the repository. It is not a release artifact, CI fixture, issue
179
+ attachment or proof that alpha.128 is installed, loaded, collecting now, or
180
+ representative of current behavior. Do not publish that report or its counts
181
+ without separate sharing approval. Synthetic tests never read production
182
+ storage. There is no scheduler, automatic analysis, upload, issue creation,
183
+ repair or live-profile mutation in this milestone.
184
+
185
+ ## Archived phase-0 qualification and proposals
186
+
187
+ The remaining sections preserve the reviewed phase-0 inventory. Their proposed
188
+ limits and future-tense acceptance steps are historical, superseded by the
189
+ runtime contract above; they do not enable any additional collection or schedule.
190
+
191
+ ## Decision and evidence levels
192
+
193
+ Start with account operations, identity/Checking and compaction. Keep Client
194
+ and Host populations separate. Use fixed aggregates, never a raw error archive.
195
+ Do not add an end-to-end claim by matching clocks, account IDs or Session IDs.
196
+ Do not count a summary request or background-job admission as recovery success.
197
+
198
+ | Seam | Evidence now | Decision | Remaining qualification |
199
+ | --- | --- | --- | --- |
200
+ | Plugin Client wrappers and Host controllers | Direct current implementation inspection | Instrument plugin-owned entry/settlement only in phase 1 | Exact Client/Host gateway tests, teardown and decoding failures |
201
+ | Native `session/event` compaction brackets/checkpoints | Existing plugin listener and public event contract | Count native bracket outcomes; inspect only event type and fixed ownership fields in memory | Matching commit followed by end; cancelled/failed/no-op and interrupted brackets |
202
+ | Published basic compaction `summarize` subclass hook | Existing selected plugin engine | Separate logical bracket from physical summary attempts | Segment failures, cancellation, call caps and selected-engine coverage |
203
+ | Storage domain public root exports | Exact official source/manifest inspection | Candidate for phase 2, not selected or activated | Published artifacts, actual backend route, restart/durability/clear/corruption and capacity tests |
204
+ | Native Schedule | Exact official source/manifest inspection | Not a generic collector callback or approved daily-analysis bridge | Explicit task owner/delivery, local read seam, offline/missed execution and stop semantics |
205
+ | Copilot App automation | Available coding-session tooling, not a Host data seam | Do not configure a task in phase 0 | Prove approved aggregate access in the selected execution environment |
206
+
207
+ Source inspection proves API shape, not loaded services, backend paths, package
208
+ installation or live health. The copied qualification tree has no Git metadata;
209
+ five inspected files were compared byte-for-byte through GitHub with the exact
210
+ official commit and matched. No Core or dependency file was changed or executed.
211
+ The public root package manifests declare storage-domain and schedule exports;
212
+ this phase did not qualify their published artifact bytes or a live backend.
213
+
214
+ ## Observation inventory
215
+
216
+ Paths below identify implementation seams, not a request to read stored
217
+ credentials, conversation bodies or the values passed into those methods.
218
+
219
+ | Operation population | Safe observation position | Observable facts | Explicit blind spots |
220
+ | --- | --- | --- | --- |
221
+ | Models account action | `src/copilot-accounts-card.ts`: `run`, owned decoder and its `finally` | Client invocation, transport result, decode result, owner invalidation, state-settlement duration | State update is not paint completion; an unmounted Client cannot confirm visible completion |
222
+ | Models Checking/Identity | Same file: `hydrateIdentity`, `identityRead` lifetime | Client ensure invocation/join, result/decoding, invalidation, unsettled age | No proof the Host received a failed RPC; joined promise is not another physical lookup |
223
+ | Chat account selection/Checking | `src/copilot-usage-card.ts`: reads and `sessionAccount.set` | Client invocation/terminal result and stale viewed-owner rejection | Quota and identity can settle independently; do not combine their success rates |
224
+ | Global switch | `src/copilot-accounts-host.ts`: `switchAccount`; controller `@Remote switchAccount` | Host receipt, admission checks, identity/model validation stages, CAS, readback, terminal view diagnostic | Identity and model checks run concurrently; their durations are not additive; resolved Promise can contain an error view |
225
+ | Session override/inherit | `src/session-accounts-controller.ts`: `set`; `src/session-accounts-host.ts`: `set` | Host receipt, membership validation when applicable, CAS, readback, controller result | Inheritance does not necessarily perform identity/model network validation; final `get` can fail after CAS |
226
+ | Identity logical read | `src/copilot-accounts-host.ts`: `readIdentity` | Cache/cooldown/join/new-flight branch, logical result, named timeout/revocation | Successful cache read is not fresh GitHub validation; cooldown is not a new failed network attempt |
227
+ | Identity physical flight | Same method's new-flight creation and final cleanup | One started/settled flight, elapsed time, allowlisted failure | No URL, token, login, user ID, body or credential fingerprint enters diagnostics |
228
+ | Compaction logical transaction | `src/compaction-continuation.ts`: public `session/event` pattern | Start, matching compact-checkpoint commit, end, interrupted owner lifecycle | Current continuation presentation only retains selected-policy cases, not all compaction; summary event alone is not commit |
229
+ | Compaction physical summary | `src/manual-compaction-recovery.ts`: `summarizeManaged`/native `super.summarize` calls | Single versus segmented attempt, completion/error/cancellation, attempt count | Selected recovery engine only; stock/custom-engine requests must remain explicit uncovered attempts |
230
+ | Background command admission | `src/background-compaction.ts`: start and native job settlement | Admitted/duplicate/refused, job terminal state, fixed failure code | Job completion with no compactable history is no-op, not checkpoint success; reporter is not a second logical failure |
231
+ | Managed model requests (later) | `src/preview-route.ts`, `src/preview-provider.ts`: public adapter/fetch hooks | Plugin-route preparation/admission and bounded request outcomes | Native routes, provider directory and unowned transport remain gaps; inspect/fetch start is not supplier acceptance |
232
+ | Auxiliary classifier (later) | `src/auto-assessment-evidence.ts`, `src/auto-task-assessment.ts` | Existing safe preparation/adapter/text/finish milestones | Auxiliary failure is not main-answer failure; first text is not HTTP first byte or billing proof |
233
+ | Hosted search (later) | `src/traditional-search.ts`, `src/plan.ts`, `src/probe.ts` | Logical search, capability proof and final-query outcomes as distinct populations | Proof failure is not a sent user query; other providers and native directory errors remain outside this owner |
234
+
235
+ The account pilot includes global switch, Session override/inherit and identity
236
+ reads/physical flights. OAuth Add/reauthorization, removal, ordinary quota reads,
237
+ model refresh and all later request/search populations are inventoried but not
238
+ implicitly included in that first denominator. Adding them changes coverage.
239
+
240
+ ### Cross-layer association
241
+
242
+ Existing strict RPC descriptors do not carry a diagnostics correlation token.
243
+ Do not alter their signatures, infer correspondence from timestamps or persist
244
+ identifiers as a shortcut. Phase 1 can deliver independent Client and Host
245
+ counts with an explicit `correlation: unavailable` boundary. A future additive
246
+ public protocol can be reviewed separately with native gateway compatibility,
247
+ short-lived token admission and teardown tests.
248
+
249
+ Within one owner, a bounded in-memory operation handle may connect start/stage/
250
+ terminal events. Native compaction matching may use the existing Session object
251
+ and compaction ID **only** while the bracket is live. Neither those IDs, account
252
+ IDs, nor their hashes are aggregate dimensions, stored records or exports.
253
+ Discard handles on settlement/disposal; never rebuild them by reading history.
254
+
255
+ ## Proposed strict vocabulary
256
+
257
+ This is a proposed schema contract, not an installed Remote or runtime schema.
258
+ Phase 1 must implement a strict version-1 codec and rejection tests before use.
259
+ Unknown keys, arbitrary strings, non-finite/negative counters and impossible
260
+ stage transitions must be rejected without echoing the rejected input.
261
+
262
+ | Field | Allowed form / meaning |
263
+ | --- | --- |
264
+ | `schemaVersion` | Literal `1` |
265
+ | `pluginVersion` | Locally loaded build version, bounded validated package version; never infer from installed files |
266
+ | `coverageVersion` | Literal reviewed inventory revision; changes split comparison populations |
267
+ | `layer` | `client` or `host`; do not merge rates across them |
268
+ | `operation` | `account-global-switch`, `account-session-select`, `account-session-inherit`, `identity-read`, `identity-flight`, `compaction`, `compaction-summary`, `compaction-job-admission` |
269
+ | `stage` | Operation-specific subset of `admitted`, `rpc-invoked`, `host-received`, `identity-validation`, `model-validation`, `cas`, `readback`, `response-received`, `decoded`, `client-settled`, `summary-attempt`, `checkpoint-committed`, `ended` |
270
+ | `outcome` | `success`, `no-op`, `cancelled`, `policy-rejected`, `revoked`, `environment-fault`, `failed`, `unknown` |
271
+ | `reason` | Fixed allowlist below, `none` or `unknown`; never an Error message/stack/cause serialization |
272
+ | `count`, `durationBucket`, `ageBucket` | Safe bounded integer counters and fixed bucket enum; monotonic elapsed time, not cross-process clock subtraction |
273
+ | `coverage` | Fixed flags/counters for unavailable seams, paused intervals, dropped operations, unknown results, discarded evidence, saturated counters and interrupted owners |
274
+
275
+ Initial reason candidates: `COPILOT_ACCOUNTS_BUSY`,
276
+ `COPILOT_ACCOUNTS_CONFLICT`, `COPILOT_ACCOUNTS_CHANGED`,
277
+ `COPILOT_ACCOUNTS_COMMIT_UNCERTAIN`, `COPILOT_ACCOUNTS_IDENTITY_TIMEOUT`,
278
+ `COPILOT_ACCOUNTS_IDENTITY_UNAVAILABLE`, `COPILOT_ACCOUNTS_MODELS_FAILED`,
279
+ `COPILOT_ACCOUNTS_SETTINGS_UNAVAILABLE`,
280
+ `COPILOT_SESSION_ACCOUNTS_UNAVAILABLE`,
281
+ `COPILOT_SESSION_ACCOUNTS_COMMIT_UNCERTAIN`,
282
+ `COPILOT_ACCOUNTS_IDENTITY_TLS`, `COPILOT_ACCOUNTS_IDENTITY_NETWORK`,
283
+ `COPILOT_ACCOUNTS_IDENTITY_RATE_LIMITED`,
284
+ `COPILOT_ACCOUNTS_IDENTITY_AUTH_REJECTED`,
285
+ `COPILOT_ACCOUNTS_IDENTITY_HTTP_ERROR`, `COPILOT_ACCOUNTS_IDENTITY_INVALID`,
286
+ `COPILOT_RESPONSES_REPLAY_SCOPE_MISMATCH`, `CONTEXT_WINDOW_EXCEEDED`,
287
+ `COPILOT_MANUAL_RECOVERY_FIXED_PREFIX`, `COPILOT_MANUAL_RECOVERY_INDIVISIBLE`,
288
+ `COPILOT_MANUAL_RECOVERY_CALL_LIMIT`, `COPILOT_MANUAL_RECOVERY_EMPTY_SUMMARY`,
289
+ `COPILOT_MANUAL_RECOVERY_UNBALANCED`, plus collector-owned fixed
290
+ `rpc-unavailable`, `decode-invalid`, `summary-invalid`, `storage-unavailable`,
291
+ `storage-invalid`, `storage-write-failed`, `observer-unavailable` and `unknown`.
292
+ Finalize this allowlist against the owning schemas in phase 1; a missing code
293
+ must map to unknown, never leak the original string. Bounded Error cause-chain
294
+ inspection may extract only exact known tokens, never retain the Error.
295
+
296
+ Classify from verified origin, not just a shared code: busy/CAS rejection is
297
+ normally policy-rejected; credential invalidation is revoked; user cancellation
298
+ is cancelled only with actual cancellation evidence. Missing service is an
299
+ environment fault. Identity-unavailable/models-failed are sanitized but broad:
300
+ they do not prove network, authentication or code defect. A hung promise is
301
+ unsettled, not a declared failure. Terminal observations are idempotent: a Host
302
+ failure plus Client error plus job reporter must not become three failures in
303
+ one logical population.
304
+
305
+ ## Denominators, slow operations and success
306
+
307
+ Each layer/operation starts its own denominator at actual admission. Report
308
+ started, settled by outcome, unsettled, interrupted and dropped separately.
309
+ Phase 1 should collect every eligible success and failure, not errors only.
310
+ For one uninterrupted owner, the accounting invariant is:
311
+
312
+ `started = settled + live-unsettled + interrupted`
313
+
314
+ Dropped admission is a separate lost-opportunity count, not a fake success or
315
+ terminal failure. A restart before persistence creates an unknown gap, not a
316
+ historical interrupted count of zero. Do not report failure rate from starts
317
+ while many are unfinished; show settled/started and all outcome counts.
318
+ Cancellation/policy rejection/environment faults are not folded into code
319
+ failure rate. Sum physical summary/identity attempts separately from logical
320
+ operation counts; cache hits and joined calls keep their actual semantics.
321
+
322
+ Proposed fixed duration/age bins in milliseconds:
323
+ `[0,100)`, `[100,500)`, `[500,1000)`, `[1000,5000)`,
324
+ `[5000,15000)`, `[15000,60000)`, `[60000,300000)`, `[300000,+infinity)`.
325
+ These are descriptive bins, not calibrated alerts. Monotonic clocks and bounded
326
+ owner-local age snapshots can detect pending work without cancelling it,
327
+ polling credentials or changing timeouts. UI unmount/Host teardown closes
328
+ coverage as interrupted; it must not silently declare the operation completed.
329
+
330
+ Account CAS fulfillment is not the entire operation's success: require the
331
+ existing owning readback and valid terminal view. Client state settlement still
332
+ does not prove a painted UI or the Host's inner stages.
333
+ Compaction success requires a matching **native compact-checkpoint commit and
334
+ successful native end**. An error after a commit is a distinct failed-after-
335
+ commit observation, not an assertion that no history changed. Record observed
336
+ commit independently of terminal status; no-op, prune-only reduction and unknown
337
+ commit remain distinct. Do not inspect checkpoint content, summarize the history
338
+ or rewrite native events/Usage to obtain evidence.
339
+
340
+ ## Privacy and control proposal
341
+
342
+ No prompts, assistant/tool content, replay, grants, device codes, raw responses,
343
+ settings values, identity/login, account/Session/job/message IDs or hashes, URLs,
344
+ paths, arbitrary exception text, stacks or freeform labels. Fixed stage codes
345
+ and numeric aggregates are the entire data boundary. Native source/transaction,
346
+ retry, account selection, authorization, accounting and cancellation stay owned
347
+ by their existing implementations.
348
+
349
+ Collection controls must distinguish disabled, active, paused and unavailable.
350
+ Enabling collection does not authorize persistence, export, scheduling, uploads,
351
+ paid probes or repairs. Clearing evidence is not disabling collection; pause
352
+ records a coverage break and does not fabricate normal outcomes. Phase 1 should
353
+ default off until its control surface and enable scope are reviewed. Collector
354
+ failures remain explicit fixed diagnostics without rejecting/retrying the
355
+ operation being observed or silently writing a fallback log.
356
+
357
+ The collector must bound live handles and aggregate dimensions before allocation,
358
+ report overflow/saturation, and tear down timers/listeners completely. Proposed
359
+ phase-1 limits are 128 live handles per owner and a 1 MiB encoded snapshot ceiling;
360
+ these are implementation-review candidates, not already enforced guarantees.
361
+
362
+ ## Storage qualification and proposed phase-2 policy
363
+
364
+ Public candidate:
365
+ [`@deepseek-ai/dsh-storage-domain`](https://github.com/deepseek-ai/deepseek-harness/blob/639ed015397290b3745d163aafe02ffee4aa3f84/packages/storage/storage-domain/src/index.ts).
366
+ Its root exports `defineDomain`, `domainTable`, `DomainFacility` and the `Domain`
367
+ types. `ctx.storageDomain.open(spec)` selects a configured public backend;
368
+ `table.put/update/delete` and `global.set` await backend durability before
369
+ publishing memory changes. `update` serializes one-record read/modify/write.
370
+ `close` rejects new writes and drains accepted work. The caller owns the handle.
371
+ Absent route/backend, unsupported facet or invalid medium is an error, not
372
+ permission to open an ad-hoc file or use Settings as a log store.
373
+
374
+ Important constraints: there is no documented multi-record transaction for
375
+ atomic clear; `open` loads all records; returned values are not defensive copies;
376
+ the backend determines crash durability and physical capacity. A logical byte
377
+ limit is not a filesystem quota. Use one bounded strict aggregate snapshot or
378
+ prove interruption-safe multi-record semantics before choosing a layout.
379
+ Do not enable `backup-and-skip`: it may retain corrupt records and log underlying
380
+ causes, violating the proposed no-raw-data contract. Prefer fail-closed open with
381
+ a fixed corruption diagnostic; explicit clear/reset must operate through a
382
+ qualified public seam and report whether inaccessible corrupt storage was
383
+ actually removed. Do not promise clear can repair an unopenable domain.
384
+
385
+ Proposed policy, pending backend/runtime tests and review:
386
+
387
+ - 14-day rolling UTC-hour aggregates; expiry at write/read/startup, no raw events.
388
+ - Hard proposed 2 MiB encoded snapshot and 336 hourly windows, plus a bounded
389
+ reviewed version/dimension inventory; enforce whichever bound is reached first.
390
+ Eviction and current retained bounds are explicit. Do not silently merge versions.
391
+ - Do not persist operation handles. Persist only aggregate unfinished counts by
392
+ safe stage if reviewed; after restart they are interrupted/unknown evidence,
393
+ never resumable handles or an invitation to replay work.
394
+ - Clear increments an aggregate epoch and removes only the plugin domain's data;
395
+ pause/disable/clear have independent results. Prevent accepted pre-clear writes
396
+ from resurrecting the earlier epoch. Do not reset credentials, Settings/history
397
+ or existing Auto observations.
398
+ - Missing/unavailable/corrupt storage yields evidence-unavailable, never an empty
399
+ healthy report. No fallback to files, Client storage, logs or another backend.
400
+
401
+ Stage 2 must qualify exact published artifacts and synthetic restart, backend
402
+ failure, queue saturation, bounded load, expiry, clock rollback, clear races and
403
+ corrupt-medium behavior. No storage dependency or route is added in phase 0.
404
+
405
+ ## Scheduling qualification
406
+
407
+ [`@deepseek-ai/dsh-schedule`](https://github.com/deepseek-ai/deepseek-harness/blob/639ed015397290b3745d163aafe02ffee4aa3f84/packages/schedule/schedule/src/index.ts)
408
+ is a durable reminder service with public create/list/catalog/history/update/
409
+ delete management and after/at/every/daily/weekly/cron record types. It requires
410
+ Agents, Sessions, tools, storageDomain, Session controller and persistence.
411
+ [Runtime delivery](https://github.com/deepseek-ai/deepseek-harness/blob/639ed015397290b3745d163aafe02ffee4aa3f84/packages/schedule/schedule/src/runtime.ts)
412
+ resolves an Agent, appends a reminder through `followup`, flushes native Session
413
+ persistence and records delivery. It is **not** a public arbitrary callback
414
+ scheduler, nor a promise that a coding session can read the plugin's domain.
415
+ Creating a reminder changes native inbox/history and can lead to model work:
416
+ do not use it as hidden passive monitoring.
417
+
418
+ Before stage 3, select exactly one mechanism, name its owner, approved read
419
+ endpoint/data range, local environment, timezone, missed-run policy, report
420
+ destination and stop/delete path. A Copilot App automation may start a coding
421
+ session, but must separately prove access to the intended live Host's sanitized
422
+ aggregate reader. Service presence, task admission, execution and successful
423
+ data read are four distinct facts. No timer, automation or reminder was created.
424
+
425
+ ## Phase-1 acceptance and later decisions
426
+
427
+ The next PR must implement the reviewed codec/controls and bounded in-memory
428
+ account/compaction pilot, not persistence or scheduling:
429
+
430
+ 1. Test real Client and Host gateways independently; transport/decoding/stale
431
+ result failures, cache/join/cooldown and successful denominators must differ.
432
+ 2. Test settled-once handles, concurrent stages, cancellation/revocation, slow
433
+ pending age, overflow, clock behavior, pause/clear and owner teardown.
434
+ 3. Exercise unchanged native compaction start/attempt/commit/end, no-op,
435
+ failed-after-commit, invalid summary, replay scope, capacity and cancellation.
436
+ Stock/custom engine gaps must remain visible; do not patch their internals.
437
+ 4. Test privacy with hostile errors, nested causes and arbitrary fields. Neither
438
+ failed input nor sensitive sentinel can appear in reports, logs or storage.
439
+ 5. Preserve original operation results and existing UI/native behavior; run the
440
+ complete required CI. Installation and loaded-version acceptance are separate.
441
+
442
+ Phase-2 persistence and phase-3 daily reads require separate qualification and
443
+ review. Stage 4 adds the inventoried request/classifier/search populations.
444
+ Stage 5 compares the same operations, coverage versions, loaded plugin versions,
445
+ windows and denominators. Sample sufficiency and thresholds are declared before
446
+ claiming improvement; no data remains insufficient evidence.
447
+
448
+ Owner for the next decision: the repository maintainer reviewing #347.
449
+ Trigger: approval of this boundary/schema/control proposal before phase-1
450
+ activation. Action: implement the pilot with a linked acceptance PR, or retain
451
+ unavailable seams and narrow the scope. This phase closes only #354, not #347,
452
+ and makes no claim that #346's Checking fault has been reproduced or repaired.
@@ -0,0 +1,141 @@
1
+ # Session cross-account continuation
2
+
3
+ Explicit **visible-history continuation** is a plugin-owned, lossy outgoing
4
+ Responses policy. It is not a new memory service or portable encrypted state.
5
+ Native Core and pi-ai already preserve replay signatures and reconstruct history;
6
+ their same-provider/API/model check does not establish Copilot account identity.
7
+
8
+ ## Experience
9
+
10
+ The composer **Credits/usage** pill opens a popover containing
11
+ **Visible-history continuation** only for an open Session whose effective
12
+ selection is the managed account-discovered Copilot route. Expand it and set
13
+ **Session policy → On** to explicitly enable an existing Session after
14
+ reviewing the loss disclosure. The canonical/native route does not show this
15
+ control.
16
+
17
+ ![Built Client continuation policy in the composer Credits popover](./images/copilot-session-continuation.png)
18
+
19
+ This 0.4.0 release-candidate capture uses synthetic Session state and quota;
20
+ [capture provenance](./images/copilot-current-provenance.json) records the
21
+ source commit, build and image hashes. It is not evidence of live availability.
22
+
23
+ New, unseeded Sessions created after this feature's first successful
24
+ activation default on. Activation persists a bounded creation-time policy
25
+ epoch using Settings CAS; existing Sessions, seeded histories, forks and
26
+ unknown birth evidence remain off. Manage the separate new-Session default in
27
+ **Models → GitHub Copilot → Manage → New Session continuation default**.
28
+ Changing it affects only subsequently created Sessions. Session overrides (on, off or
29
+ creation-time default) persist across account changes and restarts. No failed
30
+ request is required first and no hourly confirmation is needed.
31
+
32
+ The enabled disclosure stays compact. Details contain the disable action.
33
+ Enable/disable changes affect future turns; the active turn retains its captured
34
+ policy and account. The UI distinguishes pending changes from the active policy.
35
+ Disabling does not undo answers; it may expose the original replay rejection on
36
+ later requests. When off, switching to a different account pauses for persistent
37
+ enable, keep off, or cancel. The ordinary UI no longer offers a next-turn-only
38
+ mode. An exact replay rejection offers **Enable visible-history continuation**
39
+ with the loss disclosure and Cancel when off; this saves the same persistent
40
+ Session policy. When already on, show status and diagnostic guidance, not another
41
+ authorization. It never automatically sends or retries; native Retry is separate.
42
+ The failure-evidence lifetime does not limit the persisted policy. Retain old
43
+ one-turn consent and temporary recovery Remote contracts for compatibility only;
44
+ the normal Client does not create these authorizations.
45
+
46
+ See the [approved account-management experience and mock](./account-management-experience.md).
47
+ On the currently pinned Core, direct Models navigation is not a public Client
48
+ service. Chat omits the nonfunctional management entry rather than accessing
49
+ private stores or replacing Settings.
50
+
51
+ ## Loss and retained content
52
+
53
+ On each admitted managed Responses turn, the first normalized payload captures
54
+ fingerprints of pre-existing completed encrypted reasoning. Those exact items,
55
+ including summaries embedded inside them, are omitted from outgoing requests
56
+ throughout that turn. Newly generated reasoning remains available to subsequent
57
+ tool steps and native retries. A later turn treats the preceding turn's reasoning
58
+ as historical, even without an account change.
59
+
60
+ Visible user/assistant messages, native tool calls and result pairing, system/tool
61
+ definitions and disk history remain under their existing owners. No encrypted
62
+ content is decoded, copied to settings or converted into invented raw reasoning.
63
+ The policy does not claim which account issued an old item.
64
+
65
+ ## Public integration and safety
66
+
67
+ `github-copilot.sessionContinuation` stores strict bounded entries containing
68
+ only Session ID, policy version, explicit enabled state and consent timestamp.
69
+ Legacy entries without enabled retain their prior true meaning.
70
+ `continuationDefaultHistory` stores chronological bounded `{enabled, changedAt}`
71
+ entries, never Session content. Native SettingsForms CAS,
72
+ hidden volatile owned-field projection and readback preserve other settings and
73
+ the active plugin fiber. A separate strict `githubCopilotSessionContinuation`
74
+ Remote uses explicit Agent lookup; temporary recovery descriptors remain intact.
75
+
76
+ Public `agent/request`, `session/event` and `agent/disposed` bind an in-memory
77
+ per-turn baseline to the exact initiating Session and request signal. The
78
+ published adapter/SDK's payload hook filters the normalized payload. Only
79
+ `github-copilot-preview` Responses chat calls participate. Classifier,
80
+ unbound requests and other protocols remain native. Cold mid-turn recovery
81
+ without a first-step boundary fails explicitly.
82
+
83
+ The persistent Session policy also covers native managed Responses compaction.
84
+ An observed native `compaction/start` bracket plus exact public Agent lookup
85
+ binds each summary before dispatch. Active turns retain their captured persistent
86
+ policy; idle manual summaries capture current Session policy. Legacy next-turn-only
87
+ consent does not authorize summaries. Each segment omits all historical encrypted
88
+ reasoning and embedded summaries, while visible intermediate checkpoints remain.
89
+ Stock and separately selected recovery engines keep their existing transactions;
90
+ no additional failure-first attempt, retry loop or engine replacement is added.
91
+ End, cancellation, disposal and model/scope mismatch revoke filtering.
92
+
93
+ An additive composer notice reports actual filtering and native outcome. Success
94
+ requires a checkpoint replacement and an error-free matching native end, not
95
+ summary text alone. Failed/cancelled outcomes stay distinct; missing Host-lifetime
96
+ evidence remains unknown after restart. A public projection folds only native
97
+ bracket IDs/phase to trigger strict status reads; it stores no content or new
98
+ events. Polling occurs only while the viewed native bracket is active and stops
99
+ on settlement/unmount. When off, an exact replay-scope rejection exposes the same
100
+ loss-disclosed persistent enable control; unrelated failures do not.
101
+
102
+ `/copilot-compact visible-history` remains a one-operation alternative when the
103
+ selected engine supports it; its consent is never made persistent by the command.
104
+ See [summary recovery](./manual-compaction-recovery.md#explicit-visible-history-summary-recovery).
105
+
106
+ The filter scans at most 65536 top-level input items and 4096 reasoning items
107
+ (including duplicates). Only encrypted reasoning is hashed, with an aggregate
108
+ 64 Mi UTF-16-code-unit processing bound per invocation and 64 Ki-code-unit chunks
109
+ that yield to cancellation and turn revocation. These are plugin work bounds,
110
+ not supplier byte/token limits. Ordinary tool output, images, nested content,
111
+ system and tool definitions are neither traversed nor serialized by this filter;
112
+ there is no whole-request 16 MiB continuation cap. Failed, cancelled, overlapping
113
+ or revoked processing cannot partially initialize the baseline. Unsupported
114
+ reasoning, malformed settings, missing boundaries and exhausted work bounds
115
+ produce named diagnostics, not partial filtering or guessed success.
116
+ Credentials, account entitlement and model capabilities remain independently guarded.
117
+
118
+ Native retry normalization retains raw reference-restoration evidence but records
119
+ the final filtered bytes for exact 408 matching. No new retry loop is introduced.
120
+ Input admission still precedes payload filtering: this is not an oversized-context,
121
+ quota, upload-timeout or native context-meter repair.
122
+ Actual token pressure still uses the initiating Agent's enabled native compaction
123
+ engine and bounded recovery: commit the native summary, rebuild the request, then
124
+ apply continuation. A continuation work-bound error is not relabeled as context
125
+ overflow. Disabled automatic compaction, missing engines, cancellation, failed or
126
+ insufficient summaries remain explicit limitations, not permission to trim history.
127
+
128
+ ## Evidence limits
129
+
130
+ Synthetic account transitions, source/Host/UI tests and exact-source native
131
+ gateway/persistence fixtures are distinct from real supplier acceptance.
132
+ The native compaction fixture covers first-step overflow and continuing-step
133
+ prevention with Auto, a durable in-memory replacement, rebuilt visible history
134
+ and synthetic historical reasoning larger than 16 MiB. It uses an external-model
135
+ double; it does not prove real supplier acceptance or recovery of a live Session.
136
+ Publication, installation, loaded runtime and controlled live A-B-B-A success
137
+ must be reported separately. Enabling this product mode is not authorized by
138
+ implementing it; live existing Sessions still require explicit loss consent.
139
+ The default-on creation policy is disclosed in account controls and can be
140
+ disabled globally or overridden per Session. Source tests do not establish live
141
+ default acceptance or real cross-account transport.
@@ -21,7 +21,7 @@ Reload sees newly registered IDs.
21
21
 
22
22
  ![Actual parent-following and search controls with synthetic settings](images/copilot-search-routing.png)
23
23
 
24
- The picture uses actual published Client components in an isolated browser.
24
+ The picture uses current built Client components in an isolated synthetic browser fixture.
25
25
  It does not demonstrate a live search or loaded Desktop runtime.
26
26
 
27
27
  ## Routing and fallback
@@ -4,7 +4,7 @@
4
4
 
5
5
  ## Result and boundaries
6
6
 
7
- One Host-owned OAuth account supplies many account models. The stable managed
7
+ The selected Host-owned OAuth account supplies many account models. The stable managed
8
8
  route ID is `github-copilot-preview`, displayed as **GitHub Copilot**.
9
9
  Fresh installations need no native provider definition. Upgrades preserve
10
10
  existing `llm-pi-ai.providers.github-copilot`; two real groups can remain until
@@ -15,8 +15,9 @@ single-route enforcement; Core Edit/Delete remains. Model discovery/UI lifecycle
15
15
  is described in the [current README](../README.md#install-and-sign-in), not a
16
16
  prerequisite to removing a route.
17
17
 
18
- Keep `llm-pi-ai` and authorization mounted and preserve the sole credential
19
- record `llm-pi-ai/github-copilot`. Do not sign out, delete credentials or remove
18
+ Keep `llm-pi-ai` and authorization mounted and preserve the canonical credential
19
+ record `llm-pi-ai/github-copilot` and any independently authorized saved accounts.
20
+ Do not sign out, delete credentials or remove
20
21
  authorization to hide a group. `models: []` can mean the default catalog, not
21
22
  disabled models.
22
23
 
@@ -111,7 +112,7 @@ may recreate canonical, not a second account.
111
112
 
112
113
  新安装无需原生 provider;升级保留已有 canonical 配置,两组都是真实路由,
113
114
  不是两个账号。迁移只在明确批准后移除审核过的
114
- `llm-pi-ai.providers.github-copilot`;唯一凭据、`llm-pi-ai`、授权服务及其它配置保留。
115
+ `llm-pi-ai.providers.github-copilot`;canonical 凭据、独立授权的已保存账号、`llm-pi-ai`、授权服务及其它配置保留。
115
116
  不要退出登录、删除凭据或使用 `models: []` 来隐藏分组。
116
117
 
117
118
  先验证发布版本已实际加载,再调用无参数 `githubCopilot.migrationStatus()`。