From 5391ccdec27c33d59f50578cf301c965371c67bd Mon Sep 17 00:00:00 2001 From: Bingran You Date: Thu, 9 Apr 2026 14:02:09 -0700 Subject: [PATCH] docs: refresh released oracle observations --- .../native-install-update-and-setup-token-cli-flows.md | 9 +++++---- .../released-cli-e2e-test-set.md | 10 +++++----- 2 files changed, 10 insertions(+), 9 deletions(-) diff --git a/platform-services/native-install-update-and-setup-token-cli-flows.md b/platform-services/native-install-update-and-setup-token-cli-flows.md index 51e23d4..0d62e3a 100644 --- a/platform-services/native-install-update-and-setup-token-cli-flows.md +++ b/platform-services/native-install-update-and-setup-token-cli-flows.md @@ -76,7 +76,8 @@ Equivalent behavior should preserve: instead of forcing the user to run a second command manually - repair not short-circuiting the rest of the update flow when a newer native version is still available -- up-to-date states being explicit, while available updates end in a concrete +- up-to-date states being explicit when the shipped native build is already at + the newest observed release, while available updates still end in a concrete success summary such as `Successfully updated from X to version Y` The rebuild target is not merely "check if a version differs." It is a @@ -105,9 +106,9 @@ Equivalent behavior should preserve: - a TTY/raw-mode requirement because the flow expects interactive browser or manual-code handoff - a welcome banner before browser launch -- a warning when another auth source is already configured, clarifying that this - flow will create an additional OAuth token rather than replacing the main - session +- a warning only when another auth source is already configured, clarifying + that this flow will create an additional OAuth token rather than replacing + the main session - browser-open first, followed by a manual fallback URL and pasted-code prompt if the browser does not complete the flow quickly - the resulting token being treated as limited-scope material, not as a signal diff --git a/reconstruction-guardrails/verification-and-native-test-oracles/released-cli-e2e-test-set.md b/reconstruction-guardrails/verification-and-native-test-oracles/released-cli-e2e-test-set.md index 727ccad..e57a561 100644 --- a/reconstruction-guardrails/verification-and-native-test-oracles/released-cli-e2e-test-set.md +++ b/reconstruction-guardrails/verification-and-native-test-oracles/released-cli-e2e-test-set.md @@ -21,13 +21,13 @@ soft_links: Source-derived contracts are still the primary clean-room evidence for this tree, but they are not enough on their own for end-to-end rebuild work. A released CLI can be exercised directly, and its public runtime behavior becomes a second kind of oracle: not hidden implementation, but what a real user actually experiences. -This leaf captures that public-runtime oracle set from a local macOS run of the shipped `claude` CLI on April 9, 2026. +This leaf captures that public-runtime oracle set from a local macOS run of the shipped `claude` CLI on April 9, 2026. The current authoritative released binary on this machine reported version `2.1.98 (Claude Code)`. ## Snapshot and evidence boundary -Final authoritative evidence in this leaf comes from the native Mach-O binary at `~/.local/bin/claude`, which reported version `2.1.96 (Claude Code)`. +Final authoritative evidence in this leaf comes from the native Mach-O binary at `~/.local/bin/claude`, which now reports version `2.1.98 (Claude Code)`. -An earlier pass on this machine discovered an older `2.1.89` rewrite-backed shim in `~/.local/share/claude/versions/2.1.89`. That shim is not the right oracle for the shipped native CLI. Do not treat shim-era results as authoritative for released behavior. +Earlier same-day passes on this machine also surfaced a `2.1.96` native build and an older `2.1.89` rewrite-backed shim in `~/.local/share/claude/versions/2.1.89`. Those older identities are useful only as historical nuance. Do not treat shim-era results as authoritative for the current shipped native CLI. The test set below records only public behavior that was directly observed from the shipped CLI: @@ -62,12 +62,12 @@ The practical reconstruction rule is that provider-backed local testing on this ## Provider-backed local root-flag matrix -The table below is the practical E2E matrix for the `claude --help` root surface as observed on native `2.1.96`. +The table below is the practical E2E matrix for the `claude --help` root surface as observed on the current native release family. | Root surface | Status | Observed contract | | --- | --- | --- | | `--help` | `PASS` | Enumerated the public root flags and command families. | -| `--version` | `PASS` | Reported `2.1.96 (Claude Code)`. | +| `--version` | `PASS` | Reported `2.1.98 (Claude Code)`. | | `-p, --print` | `PASS` | Headless one-shot prompt path worked in both text and JSON envelopes. | | `--bare` | `PASS` | Reduced startup surface and still worked with Foundry-backed prompts. | | `--model` | `PASS` | `--model sonnet` resolved and returned successful prompts. |