@tokamak-private-dapps/private-state-cli 2.4.3 → 3.0.0

package/CHANGELOG.md

@@ -1,6 +1,96 @@
 # Changelog
 
-## Unreleased
+## 3.0.0 - 2026-06-16
+
+- Expanded the installed-package E2E to cover raw evidence export ZIP validation, investigator asset smoke checks,
+  installed `help commands` and `help guide --json` checks, `channel get-meta`, `help observer`, and final
+  `channel abandon-operation` behavior.
+- Updated README, command help, and investigator guidance to describe selected-network transaction submission and
+  mainnet-only evidence/key export confirmations.
+- Simplified the packaged Service Terms for ordinary users and judicial or regulatory readers by removing excessive
+  implementation and operation details while preserving self-custody, public-record, risk, liability, and Channel policy
+  disclosures.
+- Added human-mode install progress messages with step counts and elapsed time so users can see that installation is
+  still running after browser Terms acceptance.
+- Changed the browser-based install Terms page to render the packaged Markdown Terms as styled HTML instead of showing
+  raw preformatted Markdown text.
+- Changed human-mode CLI result formatting so `install` prints a concise installation summary and generic command
+  results render nested objects and arrays as readable indented text instead of raw JSON blobs.
+- Changed managed runtime installation subprocesses to stay quiet on successful installs while preserving captured
+  stdout and stderr in failure diagnostics.
+- Fixed `install --include-local-artifacts` to scan local `deployment/chain-id-*` artifacts for the selected DApp,
+  install local artifacts per chain, and let local artifacts override downloaded artifacts for the same chain.
+- Added optional `install --network <NAME>` scoping. Mainnet install and install without `--network` remain Terms-gated,
+  while Sepolia and anvil installs can run without interactive Terms acceptance for test and E2E workflows.
+- Changed consent-gated CLI commands to require Terms acceptance only for mainnet-targeted runs. Sepolia and anvil runs
+  skip Terms prompts while preserving mainnet protections.
+- Added optional `--network` selectors to wallet backup/key import commands so Sepolia and anvil restore flows can run
+  non-interactively while mainnet and network-omitted imports remain Terms-gated.
+- Updated the packaged Service Terms, public Terms, Privacy Notice, and Monitoring Packet wording for the corrected
+  Provider and Channel Provider responsibility model, including Channel-scoped observer and workspace mirror services.
+- Added Channel observer URL output to the Monitoring Packet policy snapshot generator and current
+  `the-great-first-channel` policy snapshot.
+- Changed `help observer` to require `--network` and `--channel-name`, read the selected Channel's observer URL from
+  on-chain Channel metadata, and fail clearly when no Channel observer URL is registered.
+- Marked `help observer` as requiring read-only deployment artifacts and changed stale BridgeCore ABI detection to return
+  a clear reinstall-required error instead of an internal TypeError.
+- Changed human `install` to open a nonce-protected local browser Terms page by default, require browser-based human
+  acceptance before installation proceeds, keep `--terminal-terms` as an explicit terminal fallback, print an immediate
+  acknowledgement before long-running install work begins, and persist the accepted Terms version, hash, timestamp, CLI
+  package version, acceptance source, acceptance method, and Terms coverage metadata in the install manifest.
+- Added a packaged canonical Service Terms source and deterministic Terms metadata so install flows report the current
+  `termsVersion` and `termsHash` from the same source in human and JSON modes.
+- Changed Terms-gated `install --json` to report that browser-based interactive Terms acceptance is required, include
+  Terms references, avoid installing artifacts, and provide agent guidance for Terms and safety-context explanation.
+- Added renewed Terms acceptance enforcement for terms-gated commands, including JSON-mode refusal when the current
+  Terms version or hash has not been accepted by the human user.
+- Aligned the CLI README RPC example with the recommended Ankr provider and removed chat-oriented wording from the
+  User-Controlled AI Agent guidance section.
+- Clarified human-facing Terms, README, and command-reference wording for install-time Terms acceptance, no-recovery
+  limits, machine-readable fee output, and ordinary-user terminology.
+- Strengthened `wallet get-notes --export-evidence` plaintext evidence export warnings, command help, README guidance,
+  investigator guidance, and evidence-scope documentation.
+- Added Terms section references to `help guide --json` agent guidance so User-Controlled AI Agents can read the
+  canonical Terms before advising users.
+- Included the Terms definitions section in `help guide --json` Terms references and clarified that `agents.md` readers
+  must read referenced Terms sections before advising users.
+- Changed `uninstall` so the default interactive flow preserves wallet spending-key and viewing-key files, while
+  `--include-wallet-keys` deletes every local private-state CLI file.
+- Fixed `uninstall` execution by awaiting the CLI runner entrypoint and importing the Tokamak zk-EVM runtime cache root
+  resolver used by the removal path.
+- Changed wallet viewing-key and spending-key exports to require interactive secret-bearing export confirmation before
+  writing `.key` files on mainnet. Sepolia and anvil key exports can run non-interactively for test and E2E workflows.
+- Updated human guide, command reference, README, and `agents.md` wording to align with install-time Terms acceptance and
+  interactive sensitive-action confirmations instead of per-command action-impact acknowledgement flags.
+- Changed transaction-sending command warning handling to print warning summaries without requiring
+  `--acknowledge-action-impact`.
+- Added the same non-blocking warning-summary handling to `channel exit`, and removed stale internal
+  action-impact acknowledgement checks from real-funds command validation.
+- Removed deprecated acknowledgement options from the private-state CLI e2e runner so it follows the current prompt
+  policy.
+- Revised user-facing CLI help, guide, README, fee, and agent wording to prefer "Ethereum mainnet", "Ethereum account",
+  "channel-local address", and "Join Toll" over developer shorthand except where command, option, contract, or JSON field
+  names require the original term.
+- Added masked terminal helper commands for creating private-key and wallet-secret source files before
+  `account import` and `channel join`.
+- Changed `help guide --json` to point AI agents at indexed `agents.md` setup recipes instead of embedding
+  human guidance prose in the JSON payload.
+- Reworked private-state CLI agent instructions around action-first mainnet setup, Ankr RPC recommendation,
+  and user-typed wallet secrets by default.
+- Changed human `help guide` output to use an action-first setup layout with one primary prefixed command,
+  plain-language status text, follow-up instructions, and no AI-only reference or diagnostic sections.
+- Clarified the README's LLM-agent entrypoint so agents start from `help guide --json`, follow
+  `agentGuidance.refs`, and read the referenced `agents.md` recipes.
+- Refined human `help guide` setup wording for RPC configuration, account import verification, wallet-secret masking,
+  channel creation, channel-data recovery, funding, note use, and channel exit.
+- Refined the human `help guide` private-key/channel-policy wording and clarified the README `--json` purpose for
+  User-Controlled AI Agents.
+- Added deterministic `help guide --json` coverage for post-RPC missing artifacts, missing account secrets, and missing
+  wallets before channel join.
+- Clarified human channel-join guide text so users prepare the Ethereum account for any Join Toll and gas before
+  joining.
+- Added ETA and scan-rate reporting to RPC recovery progress for channel and wallet workspace recovery, including
+  structured JSON progress details in `--json` mode.
 
 ## 2.4.3 - 2026-06-01
 
@@ -42,8 +132,8 @@
 
 - Changed `help observer` and monitoring references to use the public observer URL
   `https://observer.tonnel.io`.
-- Clarified CLI help, guide output, and README guidance so channel join tolls are paid directly
-  from the L1 wallet, while `account deposit-bridge` is only for channel liquidity.
+- Clarified CLI help, guide output, and README guidance so channel Join Tolls are paid directly
+  from the Ethereum wallet, while `account deposit-bridge` is only for channel liquidity.
 - Clarified README and `help commands` guidance so AI agents answer user gas, fee, and USD cost
   questions by running `help transaction-fees --json` instead of escalating to developers.
 
@@ -109,7 +199,7 @@
 - Fixed raw evidence spend linkage so nullifier observation transactions are used as the spend
   transition references and included with their transaction, receipt, and event evidence.
 - Redesigned the bundled investigator GUI around purpose-first disclosure requests, an interactive
-  SVG note-linkage graph, node detail overlays, and Markdown ASCII-art linkage report export.
+  SVG note-linkage graph, node detail overlays, and Markdown plain-text linkage report export.
 - Updated private-state documentation to reflect `set rpc` as the only CLI RPC configuration path
   for ordinary bridge-facing and wallet commands.
 - Split the CLI entrypoint into command dispatch modules and moved the shared runtime implementation