@vultisig/cli 2.19.10 → 2.19.14

package/CHANGELOG.md

@@ -1,5 +1,26 @@
 # @vultisig/cli
 
+## 2.19.14
+
+### Patch Changes
+
+- [#1034](https://github.com/vultisig/vultisig-sdk/pull/1034) [`6643df7`](https://github.com/vultisig/vultisig-sdk/commit/6643df76cf2ff2ffe08ca4985bcaf46289714e4f) Thanks [@neavra](https://github.com/neavra)! - fix(cli): fail closed on interactive prompts in non-TTY sessions instead of corrupting stdout. A piped/redirected stdout is the machine-output (JSON) channel, but bare `send`/`execute`/`swap`/`tokens --add`/`join`/`rujira swap`/`rujira withdraw` and the password/seedphrase prompts still rendered an inquirer prompt there and then died with a generic `UNKNOWN_ERROR`/exit 7 (or, for `completion --install`, a raw `ERR_USE_AFTER_CLOSE` readline stack trace).
+
+  The session is now treated as non-interactive whenever stdout OR stdin is not a TTY — mirroring how `--output` already defaults to `json` when stdout isn't a TTY, and closing the fund-safety gap where a piped `y` could otherwise auto-confirm a signing prompt. The fail-closed guard now lives at the shared prompt chokepoint (`src/lib/prompt.ts`), so **every** command that reaches an interactive prompt — including `import`, `export`, `verify`, and `address-book --add`, which previously slipped past the per-command confirm gates — throws a stable `CONFIRMATION_REQUIRED` code (exit 12) with a clear stderr hint _before_ any prompt is drawn (a password prompt points at `--password`/`VAULT_PASSWORD`/keyring; other prompts at the relevant value flags). All interactive prompts (including the create/import/verify/reshare/settings and interactive-shell paths) are routed to stderr so they can never land on the machine-output channel, `completion --install` and `-i` refuse to run without a TTY with a graceful message, and the rujira swap/withdraw flows now unlock the vault only after the confirmation gate. `--yes`/`--confirm`, a supplied `--password`/`VAULT_PASSWORD`/keyring credential, and real interactive TTY use are unchanged.
+
+- Updated dependencies [[`4483754`](https://github.com/vultisig/vultisig-sdk/commit/4483754748190fe25654de79fc12fba0edb73963), [`6643df7`](https://github.com/vultisig/vultisig-sdk/commit/6643df76cf2ff2ffe08ca4985bcaf46289714e4f)]:
+  - @vultisig/core-chain@2.24.3
+  - @vultisig/sdk@2.19.14
+
+## 2.19.11
+
+### Patch Changes
+
+- [#1129](https://github.com/vultisig/vultisig-sdk/pull/1129) [`5949742`](https://github.com/vultisig/vultisig-sdk/commit/59497426a238a75576e92d18023747d66c9d4e7a) Thanks [@neavra](https://github.com/neavra)! - fix(portfolio): report per-chain failures instead of silently swallowing them. The `portfolio` command now fetches each chain independently — one unreachable chain no longer fails the whole command, and a fiat-value lookup failure no longer silently drops the value. The `-o json` envelope always carries a `failures: [{ chain, stage, error }]` array (empty on full success), partial failures still exit 0, and an all-chains-failed run exits with a network error (code 3).
+
+- Updated dependencies [[`3bf18a1`](https://github.com/vultisig/vultisig-sdk/commit/3bf18a18606fd1b45d50abb562eb6c3011182d48)]:
+  - @vultisig/sdk@2.19.11
+
 ## 2.19.10
 
 ### Patch Changes

package/README.md

@@ -1134,25 +1134,63 @@ Configuration is stored in `~/.vultisig/`:
 
 ## Exit Codes
 
-| Code | Meaning                                                                            |
-| ---- | ---------------------------------------------------------------------------------- |
-| 0    | Success                                                                            |
-| 1    | Usage error (bad arguments, unknown command)                                       |
-| 2    | Authentication required                                                            |
-| 3    | Network error (retryable)                                                          |
-| 4    | Invalid input (bad chain, address, amount)                                         |
-| 5    | Resource not found (token, route)                                                  |
-| 6    | External service error (retryable)                                                 |
-| 7    | Unknown/unexpected error                                                           |
-| 8    | Broadcast succeeded but post-broadcast report failed — hash is valid, do NOT retry |
-| 9    | Duplicate broadcast refused (nothing sent) — retry with --force to override        |
-| 10   | agent ask: a fund-safety guardrail blocked the requested action                    |
-| 11   | agent ask: the model refused or asked a clarifying question (no action taken)      |
+| Code | Meaning                                                                                                                |
+| ---- | ---------------------------------------------------------------------------------------------------------------------- |
+| 0    | Success                                                                                                                |
+| 1    | Usage error (bad arguments, unknown command)                                                                           |
+| 2    | Authentication required                                                                                                |
+| 3    | Network error (retryable)                                                                                              |
+| 4    | Invalid input (bad chain, address, amount)                                                                             |
+| 5    | Resource not found (token, route)                                                                                      |
+| 6    | External service error (retryable)                                                                                     |
+| 7    | Unknown/unexpected error                                                                                               |
+| 8    | Broadcast succeeded but post-broadcast report failed — hash is valid, do NOT retry                                     |
+| 9    | Duplicate broadcast refused (nothing sent) — retry with --force to override                                            |
+| 10   | agent ask: a fund-safety guardrail blocked the requested action                                                        |
+| 11   | agent ask: the model refused or asked a clarifying question (no action taken)                                          |
+| 12   | Interactive confirmation/input required but the session is non-interactive — pass --yes/--confirm or the required flag |
 
 > These are generated from the `ExitCode` enum in `src/core/errors.ts` (the single source of
 > truth) and are covered by a doc-lint test that fails if this table drifts from the code. Run
 > `vultisig --help` for the same list.
 
+### Partial failures (`portfolio`)
+
+The `portfolio` command fetches every chain independently, so one unreachable chain no longer
+fails the whole command. The `-o json` envelope always carries a `failures` array (empty when
+everything succeeded):
+
+```jsonc
+{
+  "success": true,
+  "v": 1,
+  "data": {
+    "portfolio": { "totalValue": { ... }, "chainBalances": [ /* only the chains that loaded */ ] },
+    "currency": "usd",
+    "failures": [
+      { "chain": "Bitcoin", "stage": "balance", "error": "ECONNREFUSED btc-rpc" },
+      { "chain": "Ethereum", "stage": "value",   "error": "pricing service unavailable" }
+    ]
+  }
+}
+```
+
+- `stage: "balance"` — the balance fetch failed; the chain is omitted from `chainBalances`.
+- `stage: "value"` — the balance loaded but its fiat value did not; the chain still appears in
+  `chainBalances` (without a `value`) and is also listed here.
+- `error` is a concise single-line message — never a stack trace or filesystem path.
+
+**Partial-success exit contract:** if _some_ chains loaded, the command exits **0** and reports
+the rest under `failures`. Machine consumers should branch on `data.failures.length`, not `$?`.
+If _every_ chain fails to fetch a balance, the command exits **3** (network error, retryable).
+On the human-readable (table) output, failures are printed as `Warning:` lines below the table.
+
+> **Note on `totalValue`:** `failures` describes the per-chain _breakdown_ pass (`chainBalances`).
+> `portfolio.totalValue` is computed by an independent best-effort aggregate that includes token
+> values (not just native) and silently omits any chain/token it could not price. It is therefore
+> not guaranteed to be consistent with `chainBalances`/`failures` — treat it as an approximate
+> total, and rely on `failures` (not the total) to detect which chains had problems.
+
 ## Troubleshooting
 
 ### "No active vault" error