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.
- package/AGENTS.md +96 -16
- package/CHANGELOG.md +211 -0
- package/CONTRIBUTING.md +9 -0
- package/README.md +106 -18
- package/README.zh.md +97 -18
- package/deployment-baseline.json +227 -8
- package/docs/account-management-experience.md +143 -0
- package/docs/auto-high-cost.md +154 -0
- package/docs/auto-iteration-review.json +22 -0
- package/docs/auto-review-sampling.md +60 -0
- package/docs/auto-task-routing.md +8 -6
- package/docs/automatic-model-routing.md +33 -5
- package/docs/copilot-accounts.md +376 -0
- package/docs/copilot-compaction.md +23 -2
- package/docs/copilot-usage.md +24 -4
- package/docs/evidence-driven-iteration.md +86 -0
- package/docs/images/copilot-accounts-credits.png +0 -0
- package/docs/images/copilot-accounts.png +0 -0
- package/docs/images/copilot-current-provenance.json +39 -0
- package/docs/images/copilot-model-preferences.png +0 -0
- package/docs/images/copilot-search-routing.png +0 -0
- package/docs/images/copilot-session-continuation.png +0 -0
- package/docs/manual-compaction-recovery.md +50 -7
- package/docs/mockups/copilot-account-dropdown.html +71 -0
- package/docs/mockups/copilot-account-management.html +393 -0
- package/docs/model-compatibility-acceptance.md +34 -16
- package/docs/official-first-020-rc2.md +1 -0
- package/docs/plugin-diagnostics.md +452 -0
- package/docs/session-continuation.md +141 -0
- package/docs/session-search-routing.md +1 -1
- package/docs/single-route-migration.md +5 -4
- package/lib/client.js +4415 -846
- package/lib/client.js.map +1 -1
- package/lib/diagnostics-analysis-io.js +78 -0
- package/lib/diagnostics-analysis.js +425 -0
- package/lib/diagnostics-types-FVf6zjob.js +182 -0
- package/lib/index.js +10356 -6020
- package/lib/manual-compaction-recovery.js +163 -38
- package/lib/remote.js +593 -34
- package/lib/routed-web.js +1 -1
- package/lib/{search-routing-jzNYveyK.js → search-routing-CV2hs8as.js} +139 -139
- package/lib/types/account-dropdown.d.ts +22 -0
- package/lib/types/account-model-auth.d.ts +1 -1
- package/lib/types/authorization-code-clipboard.d.ts +4 -0
- package/lib/types/authorization-controller.d.ts +2 -0
- package/lib/types/auto-allocation-card.d.ts +9 -0
- package/lib/types/auto-allocation-evidence.d.ts +38 -0
- package/lib/types/auto-allocation.d.ts +24 -0
- package/lib/types/auto-model-host.d.ts +2 -0
- package/lib/types/auto-model-routing.d.ts +6 -2
- package/lib/types/auto-review-sampling.d.ts +6 -0
- package/lib/types/auto-task-classifier.d.ts +1 -1
- package/lib/types/background-compaction.d.ts +2 -2
- package/lib/types/client.d.ts +19 -3
- package/lib/types/compaction-continuation-ui.d.ts +29 -0
- package/lib/types/compaction-continuation.d.ts +48 -0
- package/lib/types/compaction-input-estimate.d.ts +24 -0
- package/lib/types/compaction-replay.d.ts +18 -0
- package/lib/types/composer-notice-style.d.ts +7 -0
- package/lib/types/config.d.ts +42 -2
- package/lib/types/context-evidence-ui.d.ts +18 -1
- package/lib/types/context-evidence.d.ts +152 -4
- package/lib/types/copilot-account-presentation.d.ts +7 -0
- package/lib/types/copilot-accounts-card.d.ts +43 -0
- package/lib/types/copilot-accounts-host.d.ts +93 -0
- package/lib/types/copilot-accounts-remote.d.ts +85 -0
- package/lib/types/copilot-accounts-types.d.ts +43 -0
- package/lib/types/copilot-auth.d.ts +3 -1
- package/lib/types/copilot-usage-card.d.ts +23 -0
- package/lib/types/copilot-usage-host.d.ts +9 -1
- package/lib/types/copilot-usage-remote.d.ts +2 -1
- package/lib/types/copilot-usage-types.d.ts +1 -0
- package/lib/types/diagnostics-analysis-io.d.ts +23 -0
- package/lib/types/diagnostics-analysis.d.ts +295 -0
- package/lib/types/diagnostics-card.d.ts +11 -0
- package/lib/types/diagnostics-client.d.ts +13 -0
- package/lib/types/diagnostics-collector.d.ts +36 -0
- package/lib/types/diagnostics-host.d.ts +72 -0
- package/lib/types/diagnostics-remote.d.ts +109 -0
- package/lib/types/diagnostics-types.d.ts +431 -0
- package/lib/types/manual-compaction-recovery.d.ts +18 -3
- package/lib/types/model-preferences-card.d.ts +1 -1
- package/lib/types/native-settings-style.d.ts +10 -0
- package/lib/types/parent-model-follow-card.d.ts +2 -11
- package/lib/types/preview-provider.d.ts +4 -0
- package/lib/types/preview-route.d.ts +7 -5
- package/lib/types/remote.d.ts +9 -6
- package/lib/types/replay-recovery-host.d.ts +2 -2
- package/lib/types/replay-recovery-ui.d.ts +3 -1
- package/lib/types/request-body-evidence.d.ts +3 -1
- package/lib/types/responses-replay-compat.d.ts +2 -1
- package/lib/types/responses-request-compression.d.ts +33 -0
- package/lib/types/session-accounts-controller.d.ts +20 -0
- package/lib/types/session-accounts-host.d.ts +33 -0
- package/lib/types/session-accounts-remote.d.ts +114 -0
- package/lib/types/session-accounts.d.ts +38 -0
- package/lib/types/session-continuation-host.d.ts +45 -0
- package/lib/types/session-continuation-remote.d.ts +22 -0
- package/lib/types/session-continuation-types.d.ts +62 -0
- package/lib/types/session-continuation-ui.d.ts +49 -0
- package/lib/types/session-continuation.d.ts +14 -0
- package/lib/types/turn-request-models.d.ts +55 -0
- package/lib/types/turn-selection-card.d.ts +15 -1
- package/lib/types/turn-selection-host.d.ts +23 -0
- package/lib/types/turn-selection-remote.d.ts +4 -0
- package/lib/types/turn-selection.d.ts +19 -4
- package/package.json +19 -3
- package/scripts/analyze-diagnostics.mjs +91 -0
- package/scripts/iteration-review.mjs +53 -0
package/AGENTS.md
CHANGED
|
@@ -31,7 +31,7 @@ Use native DSH tools for goals, background jobs and scoped subagents. Use local
|
|
|
31
31
|
|
|
32
32
|
`dsh-github-copilot` currently admits and targets only official DSH and Windows Desktop `0.2.0-rc.2`. Ten earlier DSH pins remain in `deployment-baseline.json` as historical evidence, not current support or CI release gates. It does not own a general Copilot chat adapter. DSH's built-in `llm-pi-ai` mount owns the GitHub Copilot provider, catalog, OAuth method and grant format, token exchange, refresh, and normal model transport.
|
|
33
33
|
|
|
34
|
-
This repository owns
|
|
34
|
+
This repository owns twelve narrow surfaces:
|
|
35
35
|
|
|
36
36
|
1. A conditional authorization-service bootstrap plus Host controller that joins DSH authorization, credentials, and settings.
|
|
37
37
|
2. A Client Models provider-card contribution with one shared account-state owner, Client-safe Remote descriptors and a nested plugin-owned model-preferences disclosure; embed in an existing configured canonical row, suppress its separate footer controller, and retain footer/old-Core section fallback when no such row is mounted.
|
|
@@ -39,11 +39,55 @@ This repository owns eleven narrow surfaces:
|
|
|
39
39
|
4. Preserve intentional absence of `llm-pi-ai.providers.github-copilot`; reconcile only existing legacy profiles and verified ownership journals. Fresh installations use the single account-discovered route; existing profiles require explicit migration, never silent removal.
|
|
40
40
|
5. Direct provider-hosted search using the same Host-side credential lifecycle.
|
|
41
41
|
6. Provider-scoped tool-schema compatibility for Copilot payload behaviors; Core remains the tool and execution owner.
|
|
42
|
-
7. A bounded account-discovery route that supplies validated endpoint/capability metadata to the published native adapter, plus exact plugin-owned model exclusions enforced across its directory, Auto pools and direct admission, without maintaining model-ID routing rules or changing Core's catalog.
|
|
42
|
+
7. A bounded account-discovery route that supplies validated endpoint/capability metadata to the published native adapter, including explicit managed Responses sampling compatibility, plus exact plugin-owned model exclusions enforced across its directory, Auto pools and direct admission, without maintaining model-ID routing rules or changing Core's catalog.
|
|
43
43
|
8. Optional, provider-scoped Chat presentation for completed empty reasoning disclosures; durable content and encrypted replay metadata remain Core-owned.
|
|
44
44
|
9. Compatibility-only policy restoration and request recovery for existing dedicated planner/executor Sessions. Model roles UI, settings writes and new dedicated roots are retired (#158); never reintroduce them or migrate histories implicitly.
|
|
45
|
-
10. Managed-route estimated input/output admission, optional exact-request pressure signalling through official compaction recovery,
|
|
46
|
-
11. Optional account quota reads and a session-scoped composer usage control
|
|
45
|
+
10. Managed-route estimated input/output admission, optional exact-request pressure signalling through official compaction recovery, supported low summary effort only when no effort is already resolved, and default-off lossless gzip for explicitly opted-in account-proven HTTP Responses requests. See [compaction budgets](./docs/copilot-compaction.md). Preserve truthful capacities, requested output caps, native errors/replay and transaction ownership; no hidden history trimming, wire chunking, competing compaction service or automatic model switch. The separately selected, single-service [recovery engine](./docs/manual-compaction-recovery.md) defaults to automatic segmented recovery for known oversized summaries or one typed native capacity failure through the public subclass hook. Preserve `automaticRecovery: false`, native `auto: false`, scopes and custom engines; at most 16 physical calls including a failed first attempt, no recursive retry or 408/network/auth/quota fallback. Installation alone never selects or migrates an engine.
|
|
46
|
+
11. Optional account quota reads and a session-scoped composer usage control
|
|
47
|
+
through public credentials, strict Remotes and additive slots. Keep account
|
|
48
|
+
billing-cycle credits separate from context tokens and session cost. Missing,
|
|
49
|
+
pooled, legacy and stale data must retain their real semantics; never infer credits
|
|
50
|
+
from token estimates or account deltas. A separate plugin-owned historical context
|
|
51
|
+
projection may diagnose concrete failed-zero or invalid-sample incidents; ordinary
|
|
52
|
+
request progress and absent fresh sampling stay quiet. Never filter shared native
|
|
53
|
+
usage, replace Core's meter, infer current occupancy/percentages or rewrite history.
|
|
54
|
+
See [quota boundaries](./docs/copilot-usage.md).
|
|
55
|
+
12. Default-off local account/Checking/compaction diagnostics through public storage-domain and strict additive Remotes. Keep fixed-dimension Client/Host and logical/physical populations separate, with bounded handles/rows, 14-day retention, explicit loss/uncertainty and native commit evidence. Isolate the home-wide backend by public profile context, never account/Session identity; one writer per profile is required. No content, raw errors, identities/hashes, credential/history reads, ad-hoc files, Settings logs, upload, daily task or automatic repair. Enable/pause uses only the hidden volatile `diagnosticsEnabled` CAS leaf; clear is a separately confirmed epoch fence. Missing/corrupt storage remains unavailable, never healthy empty data. Publication does not enable collection. See [local diagnostics](./docs/plugin-diagnostics.md).
|
|
56
|
+
|
|
57
|
+
## Session-policy compaction replay consent (#349, #352)
|
|
58
|
+
|
|
59
|
+
`/copilot-compact visible-history` explicitly authorizes one lossy managed Responses
|
|
60
|
+
summary operation through the separately selected recovery engine. Scope consent
|
|
61
|
+
to the native summarizer signal, Session and account-proven model protocol; revoke
|
|
62
|
+
on settlement/cancellation/teardown. The persistent visible-history Session policy
|
|
63
|
+
also authorizes managed Responses automatic/manual summaries inside an observed
|
|
64
|
+
native compaction bracket. Freeze the active turn's persistent policy, or capture
|
|
65
|
+
the idle Session policy before summary dispatch; legacy next-turn-only consent is
|
|
66
|
+
not summary consent. Bind native Agent/Session, summary model and exact signals.
|
|
67
|
+
Remove historical reasoning and its embedded
|
|
68
|
+
summaries only from outgoing payloads, preserve visible tool relations and native
|
|
69
|
+
source/transaction ownership. Other protocols stay native; explicit one-time
|
|
70
|
+
commands still fail for unsupported routes. Never silently enable a disabled or
|
|
71
|
+
unknown persistent policy. Show actual filtering and native commit/failure/cancel
|
|
72
|
+
status through strict bounded ephemeral evidence and an additive composer notice;
|
|
73
|
+
only a verified replay-scope failure offers the same enable control when off.
|
|
74
|
+
Safe background diagnostics use bounded Error cause chains and fixed allowlisted
|
|
75
|
+
codes, never raw errors. Installation alone neither selects nor invokes recovery.
|
|
76
|
+
See `src/compaction-replay.ts` and `docs/manual-compaction-recovery.md` (#349).
|
|
77
|
+
|
|
78
|
+
## Approved multi-account ownership extension (#297)
|
|
79
|
+
|
|
80
|
+
Identity presentation uses additive `ensureIdentity` Remotes, including explicit-Agent Session lookup. Keep `get` metadata-only; visible surfaces may ensure on their sixty-second cadence. Fresh ten-minute cache avoids requests, same-account consumers join a bounded flight, and failures cool down thirty seconds unless explicitly refreshed. Credential notifications revoke pending work even when token bytes are unchanged; no late cache publication. Quota may settle independently only with account-coherent metadata. Presentation cache never replaces fresh identity validation for switch, reauthorization or duplicate checks. Preserve strict sanitized diagnostics and Session/turn ownership (#309).
|
|
81
|
+
|
|
82
|
+
Account display names survive freshness expiry in bounded Host memory, never credential invalidation or failed verification. Visible account lists may ensure missing configured names without activating accounts; no disk name cache, credential copying or identity-based authorization. Turn admission captures the first verified name for the frozen account. A missing name may be filled by one existing nonforcing account-bound lookup only during that exact active, uncancelled turn/signal; never after turn/end/disposal. Evidence still requires actual native delivery and never reconstructs completed history (#322).
|
|
83
|
+
|
|
84
|
+
The user approved independently authorized account records in the existing DSH credentials service. The canonical `llm-pi-ai/github-copilot` record remains unchanged as the compatibility account. Additional plugin-owned `github-copilot/account-<opaque-id>` records must be written directly by official pi-ai OAuth through public account-bound stores; never copy/move grants, refill the canonical record, introduce a separate credential storage service, patch Core, or implement another wire/catalog. Native `Models.getAuth()` and DSH `modifyRecord()` remain the refresh and serialization owners.
|
|
85
|
+
|
|
86
|
+
Settings store only the opaque global account selector and explicit `sessionAccounts` rows. Missing/invalid selectors or records fail explicitly, without another-account fallback. Models manages reauthorization/removal and the global default; Credits may select an already authorized account for this Session's subsequent turns, restore global inheritance or run add-only authorization without selecting it (#305/#318). An explicit choice equal to the default remains explicit. Freeze account ownership before Auto assessment and request admission, holding mutation fences until native `turn/end`/disposal; later Session/default changes never redirect running steps, retries or search. Account-local credentials, metadata, quota, proof cancellation and native adapter composition must remain isolated. Keep one public adapter/controller registration, with no second wire/catalog or Core changes. Reauthorization/removal of a pinned record stays blocked; global-default selection may continue independently.
|
|
87
|
+
|
|
88
|
+
Account selection shows the current account and a compact Switch dropdown with no account search, never a flat unbounded account list. Mark the current choice, scroll long lists internally and place Add at the bottom. Saved-account reauthorization/removal remains in a separate management disclosure. One Models Manage disclosure contains account management and cross-account exact-ID model preferences. Continuation settings belong beside account controls. Persist the initial default-on epoch and later global changes with path-level CAS; only unseeded Sessions born after a known epoch inherit it, never retroactively enroll existing/unknown/forked histories. Normal Client continuation uses only the persistent Session policy; no next-turn-only or temporary-recovery duration choices. Explicit-off account changes require persistent-enable/keep-off/cancel choices, then fresh account revision readback before CAS. Exact replay failures offer loss-disclosed persistent Enable/Cancel when off, diagnostics without repeated authorization when on. Missing policy evidence fails explicitly, never a temporary authorization fallback. Retain old next-turn/recovery strict Remote contracts and Host admission for compatibility; failure-evidence expiry does not disable the persistent policy. No automatic send/retry, history writes, private Settings navigation or inferred encrypted account origin. The approved synthetic mock and current navigation limit are documented in [account management](./docs/account-management-experience.md).
|
|
89
|
+
|
|
90
|
+
Use hidden volatile `sessionAccounts` with path-level SettingsForms CAS, readback and restart evidence. Session account Remotes use explicit viewed IDs, native Agent lookup and strict codecs, never automatic Client scope. Credits uses that Session's account for identity/quota, rejecting mismatched or late results. Historical Account is an additive assistant-actions disclosure beside native Usage, captured only after a real ordinary native chunk; bounded Host-lifetime evidence becomes unknown after restart/eviction and is never reconstructed from today's settings. No new durable events, history/default/model/Usage writes, automatic account fallback, replay deletion or message retry. See [the requirements and acceptance gates](./docs/copilot-accounts.md).
|
|
47
91
|
|
|
48
92
|
## Retired model-role compatibility boundary (#158)
|
|
49
93
|
|
|
@@ -55,14 +99,23 @@ Alpha.25 admits native `subagent/descriptor` v3, already v3 in the retained rc.1
|
|
|
55
99
|
|
|
56
100
|
- `src/index.ts`: authorization bootstrap, dependency-gated Host entry, settings registration, listener, and `ctx.web` provider composition.
|
|
57
101
|
- `src/authorization-controller.ts`: sign-in/status/sign-out and route mutation.
|
|
58
|
-
- `src/copilot-grant.ts`, `src/copilot-auth.ts`: strict grant normalization and narrow pi-ai `CredentialStore` adapter
|
|
102
|
+
- `src/copilot-grant.ts`, `src/copilot-auth.ts`: strict grant normalization and narrow pi-ai `CredentialStore` adapter; canonical compatibility and explicitly selected account records follow the account-binding contract.
|
|
59
103
|
- `src/client.ts`: `settings.models.provider-card` account UI, a `plugins.bundle.config` provider-routing page with Models/footer/section fallback, and independently managed optional Chat integration.
|
|
60
104
|
- `src/web-search-routing-config.ts`, `src/web-search-routing-card.ts`: plugin-owned cross-provider live routing settings and the plugin-detail control surface. The user-facing page exposes only primary and fallback providers. Routing is stored directly under the existing `github-copilot` settings namespace at `searchRouting`, mutating both leaves via nested paths `['searchRouting', 'searchProvider']` and `['searchRouting', 'defaultSearchProvider']` in one CAS call; legacy separate routing namespaces are read as fallback only. Revisions coordinate between routing and legacy `searchModel` resets so subsequent edits do not false-conflict. Hosted-search safety settings remain schema-owned rather than a routine tuning form. The Models fallback retains routing only.
|
|
61
105
|
- `src/reasoning-presentation.ts`: guarded native Chat delegation and historical Copilot provenance; filters temporary view props only, never messages, signatures, replay indexes or usage.
|
|
62
106
|
- `src/remote.ts`: Typert Remote contribution. Never add credential payloads here.
|
|
63
|
-
- `src/copilot-usage-host.ts`, `src/copilot-usage-remote.ts`, `src/copilot-usage-card.ts`, `src/copilot-usage-ui.ts`: bounded Host quota snapshots, an independent strict namespace and reversible composer presentation. No real account fetches in tests, private Client store/DOM injection,
|
|
64
|
-
- `src/context-usage.ts`, `src/context-evidence.ts`,
|
|
107
|
+
- `src/copilot-usage-host.ts`, `src/copilot-usage-remote.ts`, `src/copilot-usage-card.ts`, `src/copilot-usage-ui.ts`: bounded Host quota snapshots, an independent strict namespace and reversible composer presentation. No real account fetches in tests, private Client store/DOM injection, quota-specific credential owner or general model wire.
|
|
108
|
+
- `src/context-usage.ts`, `src/context-evidence.ts`,
|
|
109
|
+
`src/context-evidence-ui.ts`: diagnostic-only failed-zero/invalid-sample
|
|
110
|
+
incidents and bounded numeric historical evidence through a plugin-owned public
|
|
111
|
+
projection and additive composer disclosure. Keep ordinary progress quiet;
|
|
112
|
+
dismissal is bounded Client presentation keyed by Session/incident, and a valid
|
|
113
|
+
applicable sample clears the incident. Never filter native usage chunks to
|
|
114
|
+
protect context readings: shared samples also feed Core's turn and cumulative
|
|
115
|
+
accounting. Historical counts never become current occupancy; Core pressure,
|
|
116
|
+
billing and durable events stay under Core ownership.
|
|
65
117
|
- `src/current-provider.ts`: selected DSH route plus installed pi-ai catalog facts.
|
|
118
|
+
- `src/account-model-catalog.ts`, `src/preview-provider.ts`: validated account endpoint/capability descriptors and published adapter composition. `responsesOmitTemperature` is an explicit default-off compatibility override for every managed Responses request; no supplier capability metadata is treated as temperature evidence.
|
|
66
119
|
- `src/temporary-models.ts`: exact, account-gated corrections with semantic protocol/capability retirement.
|
|
67
120
|
- `src/model-protocol.ts`: explicit Core capability detection and conservative legacy fallback.
|
|
68
121
|
- `src/responses-reasoning.ts`, `src/responses-reasoning-text.ts`: selected-model effort mapping and public summary assembly.
|
|
@@ -73,8 +126,9 @@ Alpha.25 admits native `subagent/descriptor` v3, already v3 in the retained rc.1
|
|
|
73
126
|
- `src/wire.ts`, `src/wire-anthropic.ts`: inline hosted-search streaming.
|
|
74
127
|
- `src/traditional-search.ts`: `github-copilot-hosted` `ctx.web` provider.
|
|
75
128
|
- `src/serialize.ts`, `src/sse.ts`, `src/failure.ts`: protocol conversion and bounded error handling.
|
|
76
|
-
- `src/response-error-body.ts`, `src/request-body-timeout.ts`, `src/request-upload-evidence.ts`: bounded clone-only HTTP error evidence and safe managed upload-timeout guidance. Request-scoped public Node diagnostic events supply numeric local body-write/header/ALPN evidence only after an exact verified `408/user_request_timeout`; absent events, multiple native requests and work limits stay explicit. Always dispose subscriptions when Fetch settles, without changing its arguments, Response/errors, dispatcher, proxy or retry owner. Local body-write completion is not upload duration, kernel ACK, supplier receipt or execution; its absence does not prove incomplete upload. Native failure classification/retry metadata stay unchanged; final JSON bytes are not tokens or a payload limit.
|
|
77
|
-
- `src/request-
|
|
129
|
+
- `src/response-error-body.ts`, `src/request-body-timeout.ts`, `src/request-upload-evidence.ts`: bounded clone-only HTTP error evidence and safe managed upload-timeout guidance. Request-scoped public Node diagnostic events supply numeric local body-write/header/ALPN evidence only after an exact verified `408/user_request_timeout`; absent events, multiple native requests and work limits stay explicit. Always dispose subscriptions when Fetch settles, without changing its arguments, Response/errors, dispatcher, proxy or retry owner. Local body-write completion is not upload duration, kernel ACK, supplier receipt or execution; its absence does not prove incomplete upload. Native failure classification/retry metadata stay unchanged; final JSON bytes are not tokens or a payload limit. A separate explicit opt-in compression setting is default-off and never activated by a 408; never infer overflow, auto-compact, trim history, switch models or add retries from a 408.
|
|
130
|
+
- `src/responses-request-compression.ts`: asynchronous `node:zlib` gzip at `Z_BEST_SPEED` only for exact account-proven official Copilot HTTPS `/responses` JSON POSTs through the existing public Fetch seam. `github-copilot.responsesRequestCompression` defaults false; enforce the 256 KiB minimum, 32 MiB compression-work bound and both 5%/4 KiB savings thresholds. Skips retain the original request with a fixed reason, and cancellation/account/deadline fences run before dispatch. No custom Fetch, other protocol, explicit `auto`/WebSocket, DNS/proxy/dispatcher changes, context-admission or replay changes, resend after an encoding rejection, or new retry. Gzip is wire encoding, not compaction or proof of supplier receipt.
|
|
131
|
+
- `src/request-body-evidence.ts`: verified-408-only, bounded numeric original-JSON span accounting through the existing fetch hook. Conversation/tool definitions/top-level system/residual partition total bytes; structural image and opaque replay spans are disjoint conversation subsets, never decoded contents. Unknown/limited evidence stays explicit. Fetch-to-header time excludes clone inspection and is not upload duration. With gzip, report original JSON composition and prepared HTTP-body bytes separately; neither establishes delivered bytes. No success-path analysis, retained body, automatic mitigation or supplier-limit claim.
|
|
78
132
|
- `src/copilot-stream-liveness.ts`: request-local HTTP/SSE byte-idle observation, paired with the public native adapter's bounded assistant-chunk deadline. Default five-minute byte idle and ten-minute semantic silence; no fake assistant output, private watchdog access, new retry or Core patch. Consumer think time is excluded; explicit WebSocket/auto stays native-only. `chatStreamLiveness: false` restores native-only timing. `chatMaxRequestImageBytes` retains Core's 20 MiB default; smaller explicit settings authorize only native image offload, not hidden history trimming or a guessed supplier limit. Synthetic heartbeat evidence is not production heartbeat proof.
|
|
79
133
|
- `tests/`: unit and integration evidence; mirror the source area being changed.
|
|
80
134
|
- `deployment-baseline.json`: declared machine-readable compatibility and capability evidence inventory.
|
|
@@ -87,15 +141,16 @@ Alpha.25 admits native `subagent/descriptor` v3, already v3 in the retained rc.1
|
|
|
87
141
|
|
|
88
142
|
- Do not implement a second general wire adapter or independently maintained static Copilot model catalog. The account-scoped route composes the published adapter and SDK with validated supplier metadata.
|
|
89
143
|
- Protocol and capability corrections must follow authenticated, current account metadata; new model IDs must not need new implementation tables. Native pi metadata may be reused only where it agrees with advertised endpoints and capabilities. Matching IDs alone do not prove correctness or authorize migration back into another Core catalog. Preserve old bounded ownership journals solely for verified restoration of legacy writes, never for new global protocol overrides.
|
|
144
|
+
- Native Auto review compatibility must preserve Core's frozen request and strict fail-closed verdict ownership. Automatic reviewer-only omission uses the documented public `internal/get` service-read waterfall, exact Loader root ownership and independently imported official rc.2 `apply` callback equality through same-profile `PluginPackages.packageOf` facts. Clone before native `stream` delegation only for account-proven managed Responses. Recheck identity after await and revoke on disposal; aliases, descendants, impostors and unknown package identity remain native with bounded diagnostics. Never activate a second reviewer or use private caches/registries. Missing optional reviewer packages do not block plugin startup. Ordinary concurrent Chat stays unchanged. The separate legacy `github-copilot.responsesOmitTemperature` override defaults off and affects all managed Responses only when explicitly enabled; it is not the automatic reviewer policy or a capability claim. Never infer support from thinking metadata, match reviewer prompts/models/errors, intercept canonical routes, retry/resend, migrate settings, rewrite verdicts or change permissions. See `docs/auto-review-sampling.md`.
|
|
90
145
|
- Solve protocol gaps inside the plugin using existing published extension points. Do not require a new Core capability/service or patch Core to keep a single route. The managed account-model route reuses the published adapter with one shared OAuth grant; label multiple routes honestly. Keep canonical Core models/configuration under their existing owner instead of rewriting them from the companion's pi catalog.
|
|
91
146
|
- Preserve public Responses summaries and the effective selected-model reasoning effort. Never synthesize raw reasoning replay from summaries or read/decrypt opaque replay data; assistant reasoning/replay histories delegate to Core before probing.
|
|
92
147
|
- Do not require or silently support `copilot2api`, an external gateway, a pasted GitHub token, a placeholder key, or `dsh-web-search-provider`.
|
|
93
|
-
- The credential record key is `llm-pi-ai/github-copilot
|
|
148
|
+
- The canonical compatibility credential record key is `llm-pi-ai/github-copilot`; additional account records are limited to the approved #297 contract above.
|
|
94
149
|
- OAuth credential payloads stay Host-only. Client Remote methods may expose status, notices, and errors only.
|
|
95
150
|
- Refresh must run through pi-ai `Models.getAuth()` and DSH `credentials.modifyRecord()`.
|
|
96
151
|
- Copilot OAuth grant writes must rebuild only pi-ai's documented provider fields as a fresh plain JSON object; unrelated extension values never reach DSH credential storage.
|
|
97
152
|
- Settings changes are path-level. Never replace the whole `llm-pi-ai` section or unrelated provider profiles.
|
|
98
|
-
- Sign-out deletes only the Copilot credential record and keeps route settings.
|
|
153
|
+
- Sign-out deletes only the active Copilot credential record and keeps other records, the selector and route settings. A missing selected noncanonical slot fails explicitly; only an explicit eligible Add or Switch recovers it, never automatic canonical fallback.
|
|
99
154
|
- Hosted search serves either an eligible initiating `github-copilot` / managed `github-copilot-preview` route, or provider-owned independent search selected by the web-search router. Independent search preserves a nonempty explicit `searchModel` override; otherwise the Host considers at most three eligible Responses models from current account route facts, in deterministic ID order, without a static model list. Metadata alone is not search proof, and the final user query is sent once after candidate proof, never replayed across models. Both paths require an account-available model and supported search protocol; custom inline transport remains legacy-canonical only. Never substitute the global chat default or another Session's identity. The default `probe: true` path requires successful capability proof; `probe: false` is an explicit trust override, not an implicit fallback. Any request containing a file block, including one nested in tool-result content, must bypass the custom wire through `next()` so Core retains file projection ownership.
|
|
100
155
|
- Misconfiguration and API drift fail loudly with a named missing seam. Do not fall back to process-local secrets or implicit machine state.
|
|
101
156
|
- Host, Client, and Remote package entries must stay independently buildable and exported.
|
|
@@ -141,7 +196,7 @@ The release admission, development dependency and CI targets are exact official
|
|
|
141
196
|
These pins document compatibility evidence. They do not authorize creating another controlled Core patch or making one a prerequisite for new plugin fixes.
|
|
142
197
|
|
|
143
198
|
- Models UI: `0.1.2-rc.1`, `0.1.3-alpha.1`, `0.1.5-alpha.1`, `0.1.5-alpha.2`, `0.1.5-rc.1`, `0.1.5-rc.2`, `0.1.6-alpha.1`, `0.1.6-alpha.2`, `0.2.0-rc.1` and `0.2.0-rc.2` expose `settings.models.provider-card`, keyed by settings namespace `llm-pi-ai`, to embed login/status/Refresh/Manage in a mounted configured canonical `github-copilot` row and suppress the separate footer controller. With no such row mounted, retain footer fallback; historical 0.1.1-rc.2 uses a dedicated `settings.section`. Preserve the shared account-state owner across transfer only while another eligible surface remains mounted. Unmounting the last surface or replacing declarations without overlapping mounts stops polling; a later controller reads status and separately non-forcing ensures missing/idle/stale/error/loading signed-in metadata, without replaying the old forced-login action. Manual Refresh models lives inside Manage; errors expose Retry. Opening Models is no longer guaranteed network-free, but status/details themselves remain pure. This additive slot cannot replace Core Edit/Delete: retain the native editor, while normal plugin discovery needs no manual model definitions. UI integration must not merge/remove actual canonical and `github-copilot-preview` routes or rewrite credentials, configuration, history or selection.
|
|
144
|
-
-
|
|
199
|
+
- Canonical authorization flow key: `llm-pi-ai/github-copilot`; additional account flows bind directly to their own plugin-owned record under the approved #297 contract.
|
|
145
200
|
- Authorization verification links recognize the public `window.dshDesktop.protocolVersion === 1` marker and use `_self` so Desktop's existing `will-navigate` handler opens the system browser without popup creation. Other hosts retain `_blank`; display a selectable manual URL in both cases. This does not confirm browser launch or OAuth success. Manage omits the compatibility disclosure but retains actionable legacy diagnostics and explicit repair; migration documentation and backend ownership are unchanged.
|
|
146
201
|
- Authorization service: rc.1 Core provides it; the rc.2 web/headless profiles rely on this package's runtime dependency and conditional bootstrap.
|
|
147
202
|
- Credentials: use record description/read/modify/delete APIs on the Host. Never read records in the browser.
|
|
@@ -244,13 +299,38 @@ Never say GPT-6/search works merely because settings, typecheck or a package imp
|
|
|
244
299
|
|
|
245
300
|
## Changing capabilities
|
|
246
301
|
|
|
302
|
+
Auto routing, assessment, cost weights, continuity and assistance changes must
|
|
303
|
+
first read and validate `docs/auto-iteration-review.json` using
|
|
304
|
+
`scripts/iteration-review.mjs`. Follow `docs/evidence-driven-iteration.md`:
|
|
305
|
+
missing data stays explicit, PRs record the decision and validation, and the
|
|
306
|
+
post-delivery review has an owner, sample trigger, metric and action. Never
|
|
307
|
+
upload local diagnostics automatically. See `docs/auto-high-cost.md` for the
|
|
308
|
+
orthogonal user-cost policy and separately gated advisor acceptance.
|
|
309
|
+
|
|
310
|
+
Continuation processing is reasoning-only: never recursively traverse or serialize
|
|
311
|
+
ordinary payload content to impose a whole-request limit. Preserve bounded
|
|
312
|
+
top-level item/ciphertext work, yielding cancellation and turn-revocation checks,
|
|
313
|
+
atomic baseline initialization and awaited final-byte retry evidence. Native token
|
|
314
|
+
admission and automatic-compaction transactions/rebuild stay independent; never
|
|
315
|
+
classify a continuation processing bound as context overflow.
|
|
316
|
+
|
|
317
|
+
Context evidence and replay recovery use the public full-width `conversation.input.dock`
|
|
318
|
+
above the composer, with shared native secondary typography. Do not restore long
|
|
319
|
+
notices to the nonwrapping statistics row or mutate its parent DOM. Scope errors
|
|
320
|
+
keep concise recovery guidance; bounded dispatch counts are separate Host diagnostics,
|
|
321
|
+
not proof of the rejected item, account-switch failure or recovery success.
|
|
322
|
+
Ordinary context-request progress stays quiet. Only concrete sequenced sampling
|
|
323
|
+
incidents may notify, with bounded Session/incident dismissal; missing or unknown
|
|
324
|
+
projection evidence remains a named diagnostic and never fabricated health.
|
|
325
|
+
|
|
247
326
|
Completed-turn Usage diagnostics (#259) are read-only plugin conversation data and an additive assistant-actions explanation. Official rc.2 hides its total if any attempt lacks complete accounting; the plugin's finish-only pre-dispatch pressure block can trigger this rule even after recovery. Never fabricate zero usage or replace it with successful-step partial totals. Identify local interception only from its exact recorded finish-only diagnostic, keep other missing samples and paged evidence uncertain, and suppress the explanation whenever native `tokenUsage` exists. Preserve native accounting, history, pressure and recovery; see `docs/copilot-usage.md`.
|
|
248
327
|
|
|
249
328
|
Auto task routing (#258) uses current `model_picker_category` facts, never model
|
|
250
329
|
names, effort or context capacity as a quality category. Hard eligibility/input
|
|
251
330
|
fit precede task/preference policy; unknown category and uncertain task evidence
|
|
252
|
-
remain explicit.
|
|
253
|
-
|
|
331
|
+
remain explicit. Within the first fitting category, apply positive cost weights
|
|
332
|
+
(ordinary 1, marked 0.2) and finite continuity (previous model multiplier 1.5),
|
|
333
|
+
never unconditional retention or extreme-only cost gating. Capture actual reasons once
|
|
254
334
|
per turn without durable decision events or changing Usage.
|
|
255
335
|
`autoSemanticAssessment` is default-on by explicit user rollout request (#267),
|
|
256
336
|
only for locally unknown demand; preserve explicit false opt-out. One bounded concrete
|
|
@@ -266,7 +346,7 @@ cannot supply the native gateway binding or source method discovery. Test both
|
|
|
266
346
|
actual Client and Host gateways; a synthetic RPC success cannot prove Host
|
|
267
347
|
reachability. Preserve native Agent lookup and bounded store ownership (#254).
|
|
268
348
|
|
|
269
|
-
Automatic pressure and Auto recovery availability resolve the initiating Agent's bound preset through public `agentPresets.composedPreset()` / `serviceFor(agent, 'compaction')`. A bound preset without an engine must not borrow global recovery. Only non-preset Agents use their own `agent.ctx.get('compaction')`; never scan Core's private registry or cache an engine across preset replacement. Native transactions, cancellation
|
|
349
|
+
Automatic pressure and Auto recovery availability resolve the initiating Agent's bound preset through public `agentPresets.composedPreset()` / `serviceFor(agent, 'compaction')`. A bound preset without an engine must not borrow global recovery. Only non-preset Agents use their own `agent.ctx.get('compaction')`; never scan Core's private registry or cache an engine across preset replacement. Native transactions, cancellation and retry policy remain unchanged. The selected recovery subclass uses initiating-Agent account proof and public initiator scoping for automatic/manual summaries. Multi-call results stay unmarked; omit aggregate usage after a failed initial call. No automatic profile migration or full preset replacement.
|
|
270
350
|
|
|
271
351
|
Continuing-step pressure prevention uses public `agent/pre-step` before `step/start`, only with a committed managed header, no selection notice, and known `modelSelection.pending: null`. Capacity lookup is not request estimation. Use the native surface meter and one currently bound native transaction; preserve final converted hard admission and stream-pressure fallback. First-step/pending/unknown evidence delegates. Native compaction failures propagate before attempt admission. Never claim all missing turn Usage is repaired, filter native samples, invent zero usage or rewrite historical totals. Native turn-tail accounting groups events by recorded turn, not compaction replacement sequence range.
|
|
272
352
|
|
|
@@ -283,7 +363,7 @@ Model preferences must survive the Client's owned-field projection before strict
|
|
|
283
363
|
|
|
284
364
|
Capture the traced `remote.githubCopilot` namespace once per account UI registration, shared by provider/footer/section renders. Native namespace property reads create fresh proxy identities; rereading during parent renders resets the mounted account controller and can block exclusion changes on redundant status/discovery work. Actual registration teardown and credential/connection invalidation retain their existing safety semantics.
|
|
285
365
|
|
|
286
|
-
Auto history compatibility (alpha.54, #204): official rc.2 `Session.append()` has no public ignorable-envelope option. Do not emit optional `github-copilot/auto-model-decision` events or borrow another event type; retain native real-model provenance and durable Auto selection without new durable decision recording. The assistant-actions selection footer reads bounded Host-lifetime decisions through native agent-scoped lookup and compatible historical events. Explicit captured fixed selection alone proves Manual; missing evidence is unknown, never reconstructed from today's picker.
|
|
366
|
+
Auto history compatibility (alpha.54, #204): official rc.2 `Session.append()` has no public ignorable-envelope option. Do not emit optional `github-copilot/auto-model-decision` events or borrow another event type; retain native real-model provenance and durable Auto selection without new durable decision recording. The assistant-actions selection footer reads bounded Host-lifetime decisions through native agent-scoped lookup and compatible historical events. Explicit captured fixed selection alone proves Manual; missing evidence is unknown, never reconstructed from today's picker. User-approved #310 permits model evidence only inside the existing selection dialog when native Usage lacks complete routes: successful same-turn assistant/message sources are Recorded models; public request/header snapshots inside a recorded same-turn open step are Requested model, never dispatch/execution/billing proof. Preserve multiple routes and incomplete failed-attempt attribution. Missing/invalid/paged evidence is unknown, read failures remain separate and Retry rereads the same Session/turn. Do not add a Model details/footer button, change native Usage/context metering or modify parent DOM; complete native routes avoid redundant model presentation. No closing message means no assistant-actions anchor. The independent bounded public Session projection refolds existing native events only, never emits events or inherits adjacent headers. `scripts/repair-auto-model-history.mjs` is explicit source-checkout maintenance, not a runtime hook: strict pinned official v4 validation, check-only by default, optional detached repaired copy plus byte-exact backup. It never replaces live history. Applying a copy requires separately approved stopped writers and a fresh source-hash check. See `docs/automatic-model-routing.md`.
|
|
287
367
|
|
|
288
368
|
Auto intent continuity (#293) uses the strict `githubCopilotAutoModelIntent` public Session projection of existing owner-explicit `model/selection` events. Execution headers cannot erase intent; fixed/other-provider choices replace it for future turns. Ignore copied fork/child prefixes, retain native pending and parent-follow precedence, and diagnose missing/invalid evidence. Freeze admitted Auto route configuration across late choices without writing history, settings or historical reasons. Projection disposal must remove only its own registration.
|
|
289
369
|
|
package/CHANGELOG.md
CHANGED
|
@@ -1,5 +1,216 @@
|
|
|
1
1
|
# Changelog
|
|
2
2
|
|
|
3
|
+
## 0.4.0-alpha.134 (prepared)
|
|
4
|
+
|
|
5
|
+
- Automatically omit temperature only for the authenticated official rc.2 Auto reviewer root calling an account-proven managed Responses route. Ordinary concurrent Chat and other tool listeners retain their sampling; no opt-in or settings migration is required.
|
|
6
|
+
- Qualify the public service-read facade with profile package ownership and independent exported callback equality, before cloning and delegating through the native LLM pipeline. Missing/unknown reviewer identity, aliases, descendants, other protocols and canonical routes remain native. Dispose the plugin-owned listener and revoke delayed reviewer dispatch.
|
|
7
|
+
- Execute native reviewer fixtures explicitly on unchanged source and published artifacts with exact-count/no-skip receipts. Preserve strict verdicts and fail-closed tool denial. Earlier excluded published-fixture selector claims remain retracted; synthetic qualification is not live endpoint evidence.
|
|
8
|
+
- Keep the separate all-managed-Responses override default false and preserve immutable alpha.132/alpha.133 releases. No live settings, installation or restart is included.
|
|
9
|
+
|
|
10
|
+
## 0.4.0-alpha.133 (prepared)
|
|
11
|
+
|
|
12
|
+
- Correct the #374 temperature-compatibility claim: `supports.thinking` does not establish temperature support. Replace inferred omission with the explicit, default-off `github-copilot.responsesOmitTemperature` override.
|
|
13
|
+
- When explicitly enabled, omit temperature from every account-discovered managed Responses request, independent of model metadata or native transport selection. Document the all-request scope and tradeoff; preserve native reviewer fail-closed behavior, frozen options, cancellation and one-attempt semantics. The Core-owned canonical route remains outside scope.
|
|
14
|
+
- Add default-preservation coverage for thinking true/false/absent, opt-in omission coverage for each, dynamic-config/lazy-dispatch and account/cancellation fences, plus unchanged-source native reviewer verdict cases. Synthetic fixtures do not establish supplier semantics or live endpoint success.
|
|
15
|
+
- Preserve the immutable published `0.4.0-alpha.132`; this correction uses a new release version.
|
|
16
|
+
|
|
17
|
+
## 0.4.0-alpha.132 (published; inference superseded)
|
|
18
|
+
|
|
19
|
+
- This release inferred temperature incompatibility from `capabilities.supports.thinking`. That mapping is unsupported by a verified supplier contract and must not be treated as validated or automatically capability-derived mitigation. The default-off explicit override in alpha.133 replaces it.
|
|
20
|
+
|
|
21
|
+
## 0.4.0-alpha.131 (not published; superseded)
|
|
22
|
+
|
|
23
|
+
- Earlier draft proposed temperature omission based on `capabilities.supports.thinking`; the inference is not supported and was withdrawn in alpha.133.
|
|
24
|
+
- Preserve the unused `v0.4.0-alpha.131` tag after its noncanonical manual annotation was rejected by the protected publication workflow; that version was not published to GitHub Releases or npm and is never moved, reused or repacked.
|
|
25
|
+
|
|
26
|
+
## 0.4.0-alpha.130 (prepared)
|
|
27
|
+
|
|
28
|
+
- Add explicit offline analysis for a caller-selected persisted diagnostics unit or reviewed aggregate view (#368). Bound reads, reject unsafe/changed inputs, and write strict reports only to a new explicitly selected local file.
|
|
29
|
+
- Keep version, Client/Host, operation, terminal outcome and duration populations distinct. Preserve overflow and collection gaps; do not infer failure rates, live pending state, root cause or improvement from unmatched or persisted-only evidence.
|
|
30
|
+
- Synthetic tests cover read safety and report semantics; the first descriptive report remains private local evidence. No live profile read, collection change, upload, schedule, install or restart is part of release validation.
|
|
31
|
+
|
|
32
|
+
## 0.4.0-alpha.129
|
|
33
|
+
|
|
34
|
+
- Keep ordinary post-compaction steps and requests without a fresh usage sample quiet instead of repeatedly showing context-sampling uncertainty (#367).
|
|
35
|
+
- Notify only for a concrete failed-zero or invalid-sample incident. Bind dismissal to the Session and incident across ordinary remounts/switches; a later applicable valid sample clears the incident and a newer incident can notify.
|
|
36
|
+
- Preserve the existing eight-second native-confirmed compaction result, historical-count labeling, unknown-evidence diagnostics, native usage/context meters and Core-owned compaction/accounting without replacement percentages, retries or live installation.
|
|
37
|
+
|
|
38
|
+
## 0.4.0-alpha.128
|
|
39
|
+
|
|
40
|
+
- Separate native compaction commit, subsequent ordinary-request outcome and applicable historical input sampling in one composer disclosure (#364).
|
|
41
|
+
- Preserve transient completion expiry/Close without hiding independent context uncertainty; keep native compaction history, context ring and usage accounting unchanged.
|
|
42
|
+
- Track same-step retries and compaction resumed inside an already-open native step through strict bounded public projections, without an extra request, history write or replacement percentage.
|
|
43
|
+
|
|
44
|
+
## 0.4.0-alpha.127 (prepared)
|
|
45
|
+
|
|
46
|
+
- Expire the confirmed visible-history compaction success notice eight seconds after its first Client observation; keep running loss disclosure visible and add localized, accessible Close to settled, unavailable and authorization-required results (#363).
|
|
47
|
+
- Retain expiry and presentation-only dismissal across incidental remounts and Session switches within a registration-owned 128-record bound. Scope by Session, operation and status; ignore late reads and clean up timers without changing consent, native proof/history, sending or retries.
|
|
48
|
+
- Synthetic Client lifecycle evidence is not live Desktop or provider acceptance; no installation, restart or settings enablement.
|
|
49
|
+
|
|
50
|
+
## 0.4.0-alpha.126 (prepared)
|
|
51
|
+
|
|
52
|
+
- Align plugin-owned Models, account management, model preferences, continuation disclosures, Credits details and plugin settings with the unchanged official rc.2 font family, type scale, themed neutral controls and focus states (#361).
|
|
53
|
+
- Match compact provider actions to native Edit; keep Sign out and Refresh content-sized, and wrap model actions below long names/IDs on narrow layouts. Preserve authorization, CAS, preferences, continuation, quota and native UI ownership.
|
|
54
|
+
- Isolated real-component browser comparisons cover desktop/narrow widths and light/dark themes; they do not prove live Desktop activation or authorize installation/restart.
|
|
55
|
+
|
|
56
|
+
## 0.4.0-alpha.125 (prepared)
|
|
57
|
+
|
|
58
|
+
- Add the experimental, default-off `github-copilot.responsesRequestCompression` option: account-proven managed HTTP Responses JSON may use bounded, lossless gzip at the existing public Fetch boundary (#358).
|
|
59
|
+
- Retain the original JSON for native admission, retries, replay and diagnostics; skip unsuitable bodies, preserve caller-owned Fetch and native failure handling, and never resend uncompressed or trim history.
|
|
60
|
+
- Distinguish original composition bytes from prepared gzip body bytes in verified 408 guidance. Synthetic transport evidence is not proof of delivery, provider capability or universal timeout mitigation; no live install or enablement.
|
|
61
|
+
|
|
62
|
+
## 0.4.0-alpha.124 (prepared)
|
|
63
|
+
|
|
64
|
+
- Plan compaction from model-facing content and tools instead of serializing native IDs, provenance and opaque replay envelopes, fixing metadata-driven exhaustion of the unchanged 16-call bound (#352).
|
|
65
|
+
- Retain fixed system/tools and balanced tool units when reducing the budget after one typed native capacity failure; reject an indivisible unchanged retry. Preserve actual messages, final native admission, cancellation, account proof, diagnostics and single native checkpoint commit.
|
|
66
|
+
- Add exact-source native admission/transaction regression with synthetic metadata-heavy history. Synthetic tests do not establish live recovery; no automatic Session replay, engine migration or new timeout/network retry.
|
|
67
|
+
|
|
68
|
+
## 0.4.0-alpha.123 (prepared)
|
|
69
|
+
|
|
70
|
+
- Add default-off local account/Checking/compaction diagnostics with strict fixed-dimension counts, elapsed/unfinished age bins, separate Client/Host populations and explicit cancellation, rejection, interruption and reporting gaps (#347, #356).
|
|
71
|
+
- Persist a 14-day, 4,096-row/2 MiB aggregate snapshot through the public storage-domain service, isolating profiles on the home-wide backend. Add enable/pause, separately confirmed clear and aggregate-only JSON review; preserve restart interruption, control epochs, business ownership and named storage failures.
|
|
72
|
+
- No account/Session identity, content, replay, raw errors or credentials enter aggregates. No automatic uploads, daily tasks, repair, retries, live-profile install or activation; native/custom-engine and other unwrapped paths remain explicit coverage gaps.
|
|
73
|
+
|
|
74
|
+
## 0.4.0-alpha.121 (prepared)
|
|
75
|
+
|
|
76
|
+
- Add explicit `/copilot-compact visible-history` consent for one lossy managed Responses summary operation after replay-scope rejection (#349). Preserve visible messages/tool pairs, native directive, route/cap, cancellation, source history and single transactional commit; no automatic retry, hidden settings change or live execution.
|
|
77
|
+
- Show safe allowlisted background-compaction failure reasons through bounded native Error cause chains rather than only a generic failure. Unknown causes remain explicit and private.
|
|
78
|
+
- Keep ordinary/manual/automatic compaction and chat continuation consent independent; installation alone neither selects an engine nor runs recovery.
|
|
79
|
+
|
|
80
|
+
## 0.4.0-alpha.120 (prepared)
|
|
81
|
+
|
|
82
|
+
- Let explicitly account-bound model preparation, native dispatch and subsequent turn steps continue during global selector switching, fixing `COPILOT_ACCOUNTS_BUSY` from the overly broad acquisition fence (#344).
|
|
83
|
+
- Keep unbound/global request fences, credential authorization/removal/sign-out protection, pinned-record leases, cancellation, route checks and original turn accounts. No automatic retry, message replay, fallback account, history rewrite, Core change, installation or restart.
|
|
84
|
+
- Exercise native Responses, Chat Completions and Anthropic dispatch while switch preflight is pending, then verify that the next turn adopts the confirmed new default.
|
|
85
|
+
|
|
86
|
+
## 0.4.0-alpha.119 (prepared)
|
|
87
|
+
|
|
88
|
+
- Show Models global account metadata before background identity-name hydration, keeping Switch available when metadata and route/activity evidence allow it (#342). Hydration never authorizes a switch and late results cannot undo a confirmed change.
|
|
89
|
+
- Capture the latest parent callback without rebuilding the account card lifetime, avoiding duplicate identity ensures and lost switch responses.
|
|
90
|
+
- Overlap fresh switch identity/model validation under one bounded mutation fence. Both must succeed before CAS; failures cancel and drain sibling work before releasing the fence. Preserve credential identity, membership, route/activity, revision and post-commit checks.
|
|
91
|
+
- Keep Session account/Follows global default ownership, real running turns, native directory transport and account-bound discovery unchanged. No live latency claim, automatic write retry, Core changes, installation or restart.
|
|
92
|
+
|
|
93
|
+
## 0.4.0-alpha.118 (prepared)
|
|
94
|
+
|
|
95
|
+
- Reuse confirmed model-preference CAS readback instead of describing all settings a third time for successful Exclude/Restore/High cost saves (#339).
|
|
96
|
+
- Keep pending and confirmed preference edits across equivalent parent snapshots. Account metadata can update independently without discarding a shared-preference save or restoring an old availability snapshot.
|
|
97
|
+
- Allow distinct model rows to wait in a bounded, mounted-only queue, while issuing only one narrow write at a time. Show Saving/Waiting explicitly; cancel unsent edits after unconfirmed outcomes or scope replacement and require saved-settings Retry, never automatic write replay.
|
|
98
|
+
- Native Models `llm/listProviders failed: Failed to fetch` remains a separate Host/gateway transport limitation, not evidence of forced account discovery. No Core changes, transport workaround, live latency claim, installation or restart.
|
|
99
|
+
|
|
100
|
+
## 0.4.0-alpha.117 (prepared)
|
|
101
|
+
|
|
102
|
+
- Separate confirmed Session account selection from quota loading (#337). Preserve correctly attributed quota for follow-mode-only changes without duplicate account or quota reads.
|
|
103
|
+
- Reuse the strict save response, immediately restore account controls and focus, and load a different account's quota independently. Slow or failed quota cannot undo a confirmed save; newer selections and invalidations revoke late responses.
|
|
104
|
+
- Keep continuation consent, post-consent revision reread, native CAS, account isolation and running-turn ownership unchanged. No forced model refresh, Core changes, live calls, installation or restart.
|
|
105
|
+
|
|
106
|
+
## 0.4.0-alpha.116 (prepared)
|
|
107
|
+
|
|
108
|
+
- Add immediate, cross-account High cost marking to existing model-preference rows through strict narrow Remotes and public SettingsForms CAS/readback (#333).
|
|
109
|
+
- Keep cost orthogonal to task difficulty and supplier category. Use positive weights 1/0.2 and finite previous-model multiplier 1.5 within the fitting category, preserving exclusion, capacity, fallback and admitted-turn ownership. Auxiliary classifiers separately prefer eligible unmarked Lightweight models.
|
|
110
|
+
- Expose bounded current-Session Host-lifetime allocation observations through native Agent lookup and the existing selection dialog, with candidate opportunities, conditional expected selections, actual selections and local JSON export. No automatic upload, durable history, quality/billing inference or native Usage changes.
|
|
111
|
+
- Require the previous Auto review in task plans, Agent contract and PR evidence; explicitly record missing live data and post-delivery owner/sample triggers (#332). Record controlled advisor requirements without shipping unverified extra calls.
|
|
112
|
+
- No Core changes, automatic installation, restart or live model calls for validation.
|
|
113
|
+
|
|
114
|
+
## 0.4.0-alpha.115 (prepared)
|
|
115
|
+
|
|
116
|
+
- Default automatic segmented recovery on in the explicitly selected compaction recovery engine (#330). Keep fitting summaries native; handle known oversized input or one typed summary context-limit failure inside the same Core transaction.
|
|
117
|
+
- Bound each recovery invocation to 16 physical calls including the first rejected attempt; preserve cancellation, balanced tool pairs, source/shrink checks, native retry ownership and output caps. Never escalate timeout/408/auth/quota/network failures or commit partial summaries.
|
|
118
|
+
- Preserve native `auto: false` and add `automaticRecovery: false` for automatic-segmentation opt-out. Bind summary capacity proof and dispatch to the initiating Agent's account; omit incomplete aggregate accounting after an initial failure.
|
|
119
|
+
- Document explicit same-scope engine migration and rollback without overriding custom engines, presets or live profiles. No Core changes, automatic installation or restart.
|
|
120
|
+
|
|
121
|
+
## 0.4.0-alpha.114 (prepared)
|
|
122
|
+
|
|
123
|
+
- Simplify the Chat account menu to configured accounts only. Move Follow global default into a separate checkbox and keep account addition in Models (#328).
|
|
124
|
+
- Freeze the current effective account when following is disabled; restore inheritance when enabled. Explicit account choices, including the current default, remain Session overrides.
|
|
125
|
+
- Preserve continuation consent, cancellation, revision-checked saving, running-turn ownership and quota isolation; restore focus after controls become enabled. Record the approved interactive mock and localized experience.
|
|
126
|
+
- No Core changes, automatic sends, installation or restart.
|
|
127
|
+
|
|
128
|
+
## 0.4.0-alpha.113 (prepared)
|
|
129
|
+
|
|
130
|
+
- Remove the continuation filter's whole-request 16 MiB refusal and recursive payload traversal. Large ordinary tool/image content is preserved; old encrypted reasoning is processed independently (#326).
|
|
131
|
+
- Bound reasoning count and ciphertext work, yield during hashing, and recheck request/turn cancellation before atomically establishing the baseline. Await filtering before exact native 408 retry-byte bookkeeping.
|
|
132
|
+
- Verify first-step and continuing-step native automatic compaction with Auto, durable in-memory summary replacement, fresh request rebuild and large historical reasoning. Preserve native token admission, disabled-auto/missing-engine behavior, summary failures and retry bounds; do not misclassify processing bytes as token pressure.
|
|
133
|
+
- No Core changes, automatic sends, live-history replay, installation or restart.
|
|
134
|
+
|
|
135
|
+
## 0.4.0-alpha.112 (prepared)
|
|
136
|
+
|
|
137
|
+
- Back the shared Models/Chat account dropdown's translucent native menu token with opaque Canvas. Underlying settings and quota text no longer bleed through in dark mode (#324).
|
|
138
|
+
- Replace CanvasText-based dropdown and Credits shadows with restrained dark elevation in both themes. Preserve account routing, confirmation, continuation consent, keyboard navigation and viewport bounds.
|
|
139
|
+
- Add theme regressions covering opacity and elevation; document translucent-token acceptance. No Core changes, live installation or restart.
|
|
140
|
+
|
|
141
|
+
## 0.4.0-alpha.111 (prepared)
|
|
142
|
+
|
|
143
|
+
- Retain verified saved-account display names beyond the ten-minute revalidation interval in bounded Host memory. Models and Session Credits resolve missing inactive account names without switching accounts; credential changes, removal and failed verification still invalidate names (#322).
|
|
144
|
+
- Capture identity for the frozen admitted account and non-blockingly ensure missing identity once per active turn. Preserve the first verified name, require native delivery for execution evidence, and reject late/aborted/completed-turn enrichment. Refresh the account footer at native completion.
|
|
145
|
+
- Keep existing strict Remote descriptors, credential ownership, model-access validation, history and Usage unchanged. No persistent name store, Core changes, live installation or restart.
|
|
146
|
+
|
|
147
|
+
## 0.4.0-alpha.110 (prepared)
|
|
148
|
+
|
|
149
|
+
- Replace searchable account-switch forms with one compact anchored dropdown in Models and Chat: current-choice checkmark, bounded internal scrolling and Add account footer (#320).
|
|
150
|
+
- Separate saved-account reauthorization/removal from switching. Preserve global confirmation, explicit-off continuation consent, fresh revision readback, cancellation and admitted-turn ownership; selecting never sends or retries.
|
|
151
|
+
- Record the approved dropdown mock and keyboard/mobile behavior. No Core changes, credential migration, live installation or restart.
|
|
152
|
+
- Unify normal continuation and failure guidance around the persistent Session policy. Remove duration radios and next-turn-only actions; when off, disclose loss and offer Enable/Cancel, when on, show diagnostics without repeated authorization. Existing temporary recovery contracts remain compatibility-only; their evidence expiry does not expire the persistent policy.
|
|
153
|
+
- Omit the nonfunctional Chat management entry and unavailable-navigation paragraph when no public Models navigation callback exists; keep full management in Models.
|
|
154
|
+
|
|
155
|
+
## 0.4.0-alpha.109 (prepared)
|
|
156
|
+
|
|
157
|
+
- Record the approved synthetic account-management mock and unify current-account/Switch interactions in Chat and Models. Search bounded account lists, add without selecting, and retain guarded full management inside one Models Manage disclosure with shared exact-ID model preferences (#318).
|
|
158
|
+
- Add persistent Session on/off/default policies, a creation-time global default for new unseeded Sessions, reversible next-turn consent and cancel-safe explicit-off account-switch confirmation. Existing/unknown/seeded histories are not enrolled retroactively. Authorization never sends or retries.
|
|
159
|
+
- Filter bounded pre-turn encrypted reasoning and embedded summaries through the published native Responses adapter; retain current-turn tool-loop reasoning, visible messages, native retries and durable history. Expose persistent/next-turn controls after exact replay failures; keep the native manual retry action.
|
|
160
|
+
- Preserve account/default/continuation settings revisions, dark/light native control readability, cancellation and stale-response fences. The pinned Core lacks a public Models deep link, so Chat provides manual navigation; no Core patch, credential copying, second transport or live installation.
|
|
161
|
+
|
|
162
|
+
## 0.4.0-alpha.108 (prepared)
|
|
163
|
+
|
|
164
|
+
- Align context evidence and replay recovery with native secondary typography throughout, including explanations and controls. Use the public full-width input dock above the composer instead of the non-wrapping statistics row (#315).
|
|
165
|
+
- Replace verbose replay-scope failure text with concise account/connection guidance, explicit recovery consent and a new-conversation alternative. Retain bounded sanitized dispatch counts separately in Host diagnostics; do not infer which encrypted item failed.
|
|
166
|
+
- Keep recovery default-off and lossy, with unchanged duration, expiry, native Send/Retry and stored-history semantics. No automatic replay, account reset, history deletion or Core change.
|
|
167
|
+
|
|
168
|
+
## 0.4.0-alpha.107 (prepared)
|
|
169
|
+
|
|
170
|
+
- Present an in-flight multi-account device authorization as progress with the verification URL, one-time code, copy action and cancel together; distinguish Host-confirmed post-OAuth identity/model verification from pending authorization. Adding an account never changes the global default.
|
|
171
|
+
- Keep technical account diagnostics secondary, show only supported blocker evidence, and clarify that account-information refresh is identity-only.
|
|
172
|
+
- Show turn identity only when captured for that turn and replace opaque unknown account IDs in Credits selectors with localized readable authorization labels (#313).
|
|
173
|
+
|
|
174
|
+
## 0.4.0-alpha.106 (prepared)
|
|
175
|
+
|
|
176
|
+
- Show truthful failed-turn model evidence inside the existing Auto/Manual info dialog, with selection reasons disclosed below (#310). Preserve successful same-turn models/providers and explicit incomplete failed-attempt attribution without changing native Usage.
|
|
177
|
+
- Add bounded public same-turn request-header evidence as Requested model, never dispatch, execution or billing proof. Missing/invalid evidence remains unknown; read-only Retry stays on the viewed Session/turn. No new footer button, durable events, history writes or entry for turns lacking a closing assistant-actions anchor.
|
|
178
|
+
|
|
179
|
+
## 0.4.0-alpha.105 (prepared)
|
|
180
|
+
|
|
181
|
+
- Renew expired account identity from visible Models and Credits through nonforcing, account-scoped ensure reads. Fresh ten-minute cache avoids requests; concurrent consumers share one bounded request and failed renewals retain a thirty-second cooldown (#309).
|
|
182
|
+
- Revoke pending identity on credential changes, selection changes and disposal, rejecting late results even when credentials retain the same bytes. Explicit account switching/reauthorization still validates fresh identity.
|
|
183
|
+
- Settle account-coherent Credits independently of slow identity renewal. Preserve Session account choices, running-turn pins, quota freshness semantics and metadata-only status reads.
|
|
184
|
+
- Add strict additive global/Session ensure Remotes and sanitized timeout, TLS, network, authentication, rate-limit and HTTP diagnostics. No authenticated provider calls, installation or restart are implied by synthetic evidence.
|
|
185
|
+
|
|
186
|
+
## 0.4.0-alpha.104 (prepared)
|
|
187
|
+
|
|
188
|
+
- Select an already authorized account inside Credits for this Session's subsequent turns, or follow the Models global default. Explicit same-as-default choices remain overrides; running turns keep their frozen account (#305).
|
|
189
|
+
- Isolate managed account metadata, OAuth resolution, Auto assessment, request recovery, quota and search proofs through the published native adapter. Default or Session preference changes do not redirect running steps, retries or search; pinned accounts remain protected from reauthorization/removal.
|
|
190
|
+
- Add Account beside native turn Usage using captured ordinary-request evidence. This bounded Host-lifetime evidence becomes unknown after restart or eviction; no durable events, history changes, inferred billing receipts or subagent totals.
|
|
191
|
+
- Persist only opaque Session preferences through native path-level SettingsForms CAS. Reuse existing Credits/turn popover styles and explicit Session lookup with separate strict Remotes; preserve native account compatibility and credential ownership.
|
|
192
|
+
|
|
193
|
+
## 0.4.0-alpha.103 (prepared)
|
|
194
|
+
|
|
195
|
+
- Reevaluate transient account BUSY failures from current leases, authorization and managed Agent activity on status reads instead of retaining a stale rejection after work ends (#303).
|
|
196
|
+
- Keep active work blocked and preserve missing evidence, native-route restrictions and uncertain-commit errors. Status recovery does not refresh credentials, authorize another account, switch models or write settings/history.
|
|
197
|
+
|
|
198
|
+
## 0.4.0-alpha.102 (prepared)
|
|
199
|
+
|
|
200
|
+
- Read official rc.2 SettingsForms values without requiring the retired `get()` method, fixing an unconditional `COPILOT_ACCOUNTS_EVIDENCE_INCOMPLETE` account-switching blocker (#301).
|
|
201
|
+
- Exercise the production route diagnostic in native account persistence and restart coverage. Preserve unknown-evidence rejection, managed-only eligibility, activity fencing, target validation and selector CAS; no credentials, history, model-selection or Core changes.
|
|
202
|
+
|
|
203
|
+
## 0.4.0-alpha.101 (prepared)
|
|
204
|
+
|
|
205
|
+
- Accept GitHub Enterprise Managed User login names, including underscore-separated enterprise suffixes and setup admin names, in both Host identity normalization and strict Client Remote decoding. Keep bounded JSON, numeric identity, credentials and account-switching safeguards unchanged.
|
|
206
|
+
|
|
207
|
+
## 0.4.0-alpha.100
|
|
208
|
+
|
|
209
|
+
- Add independently authorized GitHub Copilot accounts through official OAuth and the existing DSH credentials service (#297). Preserve the canonical compatibility account; inactive-account refresh remains lazy and record-bound.
|
|
210
|
+
- Manage, confirm switches, remove inactive authorizations and reauthorize the same identity in Models. Managed-only evidence, target preflight, activity fencing and path-level selector CAS prevent unsafe switches; no history, selected-model or default rewrites.
|
|
211
|
+
- Show the current GitHub identity read-only in Credits details, paired with account-bound quota evidence. Clear mounted readings during account changes and reject late responses; quota snapshots are not proof of cross-app AI-credit aggregation.
|
|
212
|
+
- Preserve existing Remotes and add a separate strict account namespace. Native gateway, adapter and synthetic on-disk restart tests complement the required full CI; no live OAuth, installation or loaded Desktop success is implied.
|
|
213
|
+
|
|
3
214
|
## 0.4.0-alpha.99 (prepared)
|
|
4
215
|
|
|
5
216
|
- Enforce Auto assessment's single monotonic deadline at preparation, native request and result boundaries even when a delayed timer has not fired (#295). Dispose the assessment timer and listener on settlement; preserve cancellation, proof revocation and unknown-demand fallback.
|
package/CONTRIBUTING.md
CHANGED
|
@@ -44,3 +44,12 @@ Prepare version alignment in the implementation PR. Important runtime or deliver
|
|
|
44
44
|
After the release-ready PR merges, the successful main CI run calls the reusable Release workflow on that exact SHA. It creates or reconciles the annotated version tag and a draft-first Release, rebuilds and verifies the package, writes and verifies `SHA256SUMS`, checks remote asset digests, and only then publishes an immutable Release. Identical reruns reconcile safely; conflicting tags, releases or bytes fail closed. A tag created by `GITHUB_TOKEN` does not trigger a second workflow, so publication occurs in the gated main workflow rather than depending on recursive tag events. Do not hand-upload a replacement or bypass failed CI.
|
|
45
45
|
|
|
46
46
|
Release-bearing work is complete only after reporting the published Release URL and verified version/commit/assets/checksum, or clearly identifying a concrete publication blocker and pending step. A merged PR or a local tarball alone is not release delivery. After publication, independently verify the Release is not a draft, its prerelease flag, exact tag/commit, asset names and downloaded SHA-256. An install command must include the target `--profile`; never overwrite a shipped preset. Compare the installed package to the downloaded asset, then separately validate activation after restart. HTTP 200 alone does not prove the new plugin loaded.
|
|
47
|
+
|
|
48
|
+
## Auto policy evidence
|
|
49
|
+
|
|
50
|
+
Before Auto routing, assessment, high-cost, continuity or assistance changes,
|
|
51
|
+
read and validate `docs/auto-iteration-review.json`. Follow
|
|
52
|
+
[`docs/evidence-driven-iteration.md`](./docs/evidence-driven-iteration.md) and
|
|
53
|
+
record the prior decision, evidence status/denominator, actual validation and
|
|
54
|
+
post-delivery owner/trigger in the PR. Missing observations are not a pass.
|
|
55
|
+
Never upload raw local exports or infer billing/quality from allocation counts.
|