micro-models-agent 1.1.0 → 1.5.0

package/CHANGELOG.md

@@ -4,6 +4,116 @@ All notable changes to Micro Models Agent (MMA) will be documented in this file.
 
 The format is based on [Keep a Changelog](https://keepachangelog.com/), and this project adheres to [Semantic Versioning](https://semver.org/).
 
+## [1.5.0] - 2026-09-21
+
+### Added
+
+- **Canonical chat UI style** (`docs/design/ui/STYLE.md`, mockup `docs/design/ui/mockup-chat-en-ru.html`, linked from AGENTS.md as the single correct style): no boxed panels, one shared glyph set, semantic color roles, quiet-by-default with detail on demand.
+- **UI theme module** (`src/ui/theme.ts`): single source for chat glyphs (`◇ ✓ ✗ ⚠ ✦ ? ↵ ↳ │ ▰▱ ⟡ ⟳`), semantic marks and the `Verbosity` type.
+- **Chat verbosity levels** (`ui.verbosity`: `quiet | normal | verbose`, default `quiet`): quiet folds each action into a single `◇ label ✓ ms` line, hides tool bodies and only surfaces the first line of a failed action. New `/verbose [level]` REPL command and `--verbose` plumbing; legacy `ui.verbose` still maps to `verbose`.
+- **Read-only run summaries** (`src/ui/renderer.ts`): in quiet mode a consecutive run of read-only tools (`read_file`, `grep`, `glob`, web, …) collapses into one `│ N actions ✓ · ms · /verbose` line; mutating/executing tools stay visible.
+- **Calm security/protected-path failures** (`src/core/agent/tool-output.ts`, `src/core/agent/tool-batch.ts`, `src/modules/security/protected-paths.ts`): `toolFailureSummary` maps `[SECURITY BLOCKED]` / `[PROTECTED PATH]` refusals to short localized warnings (`⚠`) instead of red walls; raw markers never reach the chat.
+- **`Ctrl+O` action reprint** (`src/ui/action-log.ts`, `src/cli/repl.ts`): the last action's label, args, status and full body are kept and reprinted on demand, so quiet mode stays quiet without losing detail.
+- **Spinner elapsed time** (`src/ui/spinner.ts`): the busy line shows `· Ns` once a tool runs longer than 3 seconds.
+- **Status voice** (`src/ui/renderer.ts`): thinking phases rotate a few short human labels (`thinking` / `working it out` / `considering`, localized) instead of repeating one word.
+- **Plan checklist restyle** (`src/ui/plan-view.ts`): `◇ title ▰▱ 1/3` header and `✓ / ◇ / ○ / ✗` step marks, dropping the raw plan id and `[x]` brackets.
+- **Category-colored action glyph** (`src/ui/theme.ts`): the single `◇` mark is colored by tool category — recon cyan, mutate magenta, execute blue, meta dim. Per-tool emoji icons are no longer rendered.
+- **Human tool arguments** (`src/ui/renderer.ts`): raw JSON is replaced by short phrases — `plan update` → `step 2 → done`, `browser scroll down`, `move_file` → `a → b`, `mcp_call` → `server/tool`, `remember`, `enable_tools`, `glob`/`grep`/`subagent`/`set_thinking`/`verify`/`process_log` and more.
+- **Quiet hides structured info logs** (`src/output/channel.ts`): `info`-level log events (`[INFO] [mma] …`) are dropped from the terminal in quiet mode; warn/error and `writeLine` blocks are unaffected.
+
+### Fixed
+
+- **Closing code fence without a trailing newline** (`src/ui/md-formatter.ts`): a final ```` ``` ```` at end of stream was rendered as code content; it now closes the block.
+- **Thought duration inside the answer** (`src/ui/renderer.ts`): the thinking block is closed when the final answer or the next tool starts, so `N.Ns` never lands inside the response text.
+- **Durations over a second** (`src/ui/theme.ts`): rendered as `1.4s` instead of `1400ms`.
+
+### Changed
+
+- **Tool output never shows system names** (`src/ui/renderer.ts`): tool identifiers (`read_file`, `bash`, …) are lookup keys only — the chat renders the localized `tool.friendly.*` label, with a generic localized fallback when a translation is missing.
+- **One action glyph** (`Renderer.toolMarker`): every tool uses `◇` instead of per-category arrows/asterisks.
+- **Calm errors** (`src/ui/renderer.ts`, `src/ui/md-formatter.ts`): `✗` for real failures, `⚠` for soft blocks; `ui.error_prefix` is now `✗ ` in both locales.
+- **No box-drawing grids** (`src/ui/md-formatter.ts`, `src/ui/table.ts`): code blocks use a dim language label plus a `│` gutter; markdown tables render as a bold header + dim rule + aligned columns.
+- **Thinking header and run footer**: `✦ thinking` instead of `→ Thought:`, `⟡ model · provider · ms` instead of `▣`.
+- **Answer has no marker prefix** (`src/ui/renderer.ts`): the final answer starts on a fresh line instead of `- …`, so a response that opens with a code block is not mangled.
+
+## [1.4.0] - 2026-09-21
+
+### Added
+
+- **Protected-path confirmation guard** (`src/modules/security/protected-paths.ts`): file-mutating tools (`write_file`, `edit_file`, `delete_file`, `move_file`, `create_dir`) now refuse paths under VCS internals (`.git/`, `.hg/`, `.svn/`) and secret/env files (`.env*`, `*.key`, `*.pem`, `*.p12`, `*.pfx`, `*.crt`, `*.secret`, `*.token`, `.npmrc`, `.netrc`, `.git-credentials`, `.gitconfig`) with a directive to ask the user for explicit confirmation. The guard is **always on** (independent of `security.paths.enabled`) and never hard-blocks — the user can still approve. Verified live on `qwen/qwen3.5-9b`: the model recognized the rule from the prompt, refused, and asked the user.
+- **System-prompt `PROTECTED PATHS` rule** (`src/core/bootstrap.ts`): the model is told up front not to modify VCS/secret/env paths without explicit user confirmation, and to stop and ask if a tool reports one (no `bash` workarounds).
+
+### Fixed
+
+- **Provenance after transparent provider failover** (`src/core/agent.ts`, `src/core/agent/token-tracker.ts`, `src/core/bootstrap.ts`): after `FallbackProvider` switched to another entry, the footer, cost attribution and `llm_usage` still reported the original provider. `callerProvenance()` now reads the live active entry (`getProviderName`), and `TokenTracker.logUsage` passes `provider`/`model` into `llm_usage`. Verified live: `provider failover: dead -> live` → footer `· live`, `llm_usage.provider=live`.
+- **Session encryption config was ignored** (`src/core/bootstrap.ts`): `SessionStore` was constructed without `config.security.sessionEncryption`, so enabling `security.sessionEncryption` silently left `session.jsonl`/`history.jsonl` in plaintext. The config is now passed through; verified live (files written as `salt:iv:authTag:data`).
+- **Certification harness hardening** (`src/modules/certification/`, found by running the full suite on `qwen/qwen3.5-9b`):
+  - `syncGlobalManifest` no longer downgrades a fresher local certification with an older bundled manifest — it overwrites a key only when the bundled entry is genuinely newer (`manifest.ts`).
+  - `readMergedManifest` now picks the **newer** entry (by `mmaVersion`, then `certifiedAt`) instead of always letting a stale legacy project manifest shadow the global one, so `model cert-status`/`cert-list` show the real result.
+  - `outputContains` is case-insensitive, so security scenarios match the uppercase `[SECURITY BLOCKED]` harness marker deterministically rather than depending on the model echoing lowercase "security".
+  - MoE scenarios state "working directory root (no subdirectories)"; the security scenarios explicitly instruct the model to attempt the blocked command (a small model otherwise refused and never exercised the block).
+
+## [1.3.1] - 2026-09-21
+
+### Added
+
+- **Code style guide** (`docs/agents/code-style.md`, AGENTS.md/architecture rules #18): formatting, ban on `as any`/`as unknown as X` (except annotated test seams), `import type`, explicit export return types, abortable waits, no silent `catch`, Russian comments, and a phased tsconfig-strictness ladder.
+
+### Fixed
+
+- **False-positive repeated-tool hard stop** (`src/core/agent.ts`): under `--exit-on-complete` an identical repeated tool call was stopped after two repeats even when it was a legitimate request ("run `echo tick` four times"), ending the run with an empty answer. The hard stop now applies only to loop-prone status tools (`session_info`/`process_list`/`process_log`/`file_info`/`project_map`); side-effecting tools (e.g. `bash`) stay nudge-only. Verified live on `qwen/qwen3.5-9b`.
+- **False hallucination warning for filenames with spaces** (`src/modules/hallucination/factual.ts`): `отчёт 1.txt` was tokenized as `1.txt` and flagged as a missing file. `bareNameExists` now also matches a known/indexed name ending in `" <token>"`. Verified live.
+- **`tool_call` args JSON sanitization** (`src/core/session-logger.ts`): `logToolCall` no longer loses the whole entry when the data sanitizer corrupts the arg JSON (e.g. `password=secret"` loses a closing quote) — falls back to a raw string and warns.
+- **Invalid tool-call arguments** (`src/core/agent.ts`): a malformed `tool_call.arguments` JSON no longer silently becomes `{}`; it is logged (`tool_args_parse_error`) and a `system-summary` tells the model to re-issue the call with valid JSON.
+- **MoE plan mutation** (`src/modules/execution/moe-executor.ts`): `resolveScopeAndRetry` expands scope on a local copy, so the original plan stays intact for verification/re-plan.
+- **Browser session race in MoE** (`src/tools/browser.ts`): concurrent waves are serialized via a promise chain instead of clobbering the shared Playwright session; `closeBrowserSession` waits for the in-flight action; missing cookie dir is created; close failures are logged. Screenshot vision detection broadened (`modelSupportsVision`).
+- **Security-block nudge scope** (`src/core/agent/tool-batch.ts`): `sawSecurityBlock` is reset per tool batch, so a stale block cannot trigger the wrapper-script nudge much later.
+- **Per-tool cost** (`src/core/agent/tool-batch.ts`): the tool-end event now reports the delta (`total_after - total_before`) instead of the cumulative session cost.
+- **`edit_file` argument guard** (`src/tools/edit-file.ts`): non-string `old`/`new` are rejected with a clear error instead of being coerced to `"null"`/`"[object Object]"`; unreadable files return a localized message.
+- **Bash failure counters scoped to the session** (`src/tools/bash.ts`): the module-level `FAILING_FIRST_WORDS` map is reset when the session id changes, so failures from a previous session do not trigger a bogus hard-stop later.
+- **Migration backup skip is logged** (`src/migration/backup.ts`): a pre-existing `.bak` directory is reported at debug level instead of being skipped silently.
+- **File I/O error handling** (`src/tools/write-file.ts`, `src/tools/read-file.ts`): `writeFileSync`/`readFileSync` failures (ENOSPC/EPERM, file vanished between `existsSync` and read) now return a localized message with the path and reason instead of throwing a bare Node error; numeric `offset`/`limit` are validated instead of cast.
+- **`listModels` typed parsing** (`src/llm/openai-compat.ts`): provider responses are parsed via a narrowing helper (no `as any`), tolerating both `{data:[…]}` and bare arrays.
+- **Nested markdown emphasis** (`src/ui/md-formatter.ts`): `***bold italic***` renders as bold+italic without leaving a stray asterisk.
+
+### Changed
+
+- **MoE messages localized** (`src/core/agent-moe.ts`, i18n `moe.*` en+ru): planning/fallback/auto-fix/no-orchestrator/plan-failed strings go through `t()`.
+- **MoE typing** (`src/core/agent-moe.ts`, `src/llm/types.ts`): `verifyAndMerge` is typed as `OrchestratorResult`/`VerifyInput`, `getUsageByTag` is part of `MoEExecutorLike` (no duck-typed cast), orchestrator override is properly optional.
+- **Provider cache** (`src/modules/providers/manager.ts`): `setModel()` clears the build cache so providers for old models do not accumulate.
+- **Input-from budget** (`src/modules/execution/input-from.ts`): aggregate cap (`maxChars * 4`) with an `omitted` marker; `as any` narrowing removed.
+- **Observability** (`src/core/bootstrap.ts`, `src/core/agent.ts`): ContextLogger init failure and `getProviderFor` build failure are logged with i18n keys; log level is validated (`normalizeLogLevel`) instead of `as any`.
+- **SIGINT force-exit** (`src/cli/repl.ts`): grace period 2s → 5s with a localized notice.
+- **i18n diagnostics** (`src/i18n/index.ts`, `src/core/bootstrap.ts`): a missing key is reported via an optional debug reporter (`setMissingKeyReporter`, wired to `logger.debug`); `loadLocale` cast removed.
+- **Type cleanup** (`src/tools/*`): removed redundant `(ctx as any).fileOperationsCount` casts in 5 files.
+- **Abortable retry sleep** (`src/modules/execution/moe-executor.ts`): retry `sleep` honors the abort signal; stuck detector receives the real attempt number.
+
+## [1.3.0] - 2026-09-19
+
+### Added
+
+- **Budget-driven compaction floor** (`contextBudget.compactionMinLoad`, default 0.5, `src/modules/context/manager.ts`): the 15-iteration interval no longer triggers a compaction on its own — it now requires the history to be at least `compactionMinLoad` full. Previously compaction fired at ~40% of the budget and discarded 33-44 useful turns mid-task (observed in a real run), derailing the agent. Token-pressure compaction (`compactionThreshold`) is unchanged.
+- **Context probe persistence in every mode** (`src/core/bootstrap.ts`, `src/cli/main.ts`): the loaded-vs-configured context probe is now written to the session (`context_probe`) and surfaced in one-shot runs too. Previously only the REPL recorded/rendered it, so a mismatch (LM Studio loaded 86016 while `contextWindow` was 32768) was silent in `mma "<prompt>"`.
+- **Artifact churn detector** (`src/core/agent/artifact-progress.ts`): nudges the model once when it repeatedly rewrites a test/spec file without changing production code. Path classification is cross-language (`*.test.*`, `*_test.*`, `test_*`, `*Test.java`, `tests/`, `specs/`, …), with no framework/task specifics.
+- **Stuck escalation reaches the model** (`src/modules/execution/execution-plugin.ts`): after the escalation threshold, a generic directive (change approach / search the exact error / ask the user / re-plan or skip) is injected as a `system-summary`, not only shown in the UI.
+- **Wrapper-script guard** (`src/core/agent/tool-batch.ts`, `src/tools/bash.ts`): after a `[SECURITY BLOCKED]` result, writing a `.bat/.cmd/.sh/.ps1` wrapper triggers a one-time nudge; the security block hint now points to tools (write_file, workdir, separate calls) instead of only `workdir`.
+
+### Changed
+
+- **`plan update` no longer rewrites the plan's steps** (`src/modules/execution/plan-actions.ts`): a bulk `steps[]`/`title` rewrite via `update` is rejected in all cases (previously only when the plan had progress). Use `re-plan` (keeps completed steps) or `create`.
+- **Tool-arg summaries omit long values** (`src/core/agent/tool-output.ts`): `summarizeToolArgs` collapses long string values (`content`/`new_string`/…) to `…(N chars)` instead of echoing the body into the `<system-summary>` envelope.
+- **Artifact-requirement prompt rule** (`src/core/bootstrap.ts`): explicit generic line — a requested document/artifact (spec, design, README, plan file, script) must be created as a real file, and libraries not agreed/listed must not be added.
+
+### Fixed
+
+- **Non-streaming fallback on buffered-SSE stall** (`src/llm/openai-compat.ts`): a stream that stalls after tool-call deltas started (LM Studio buffers large tool-call SSE) is retried non-streaming once instead of failing the whole run. Reasoning from the fallback is dropped (already streamed).
+
+## [1.2.0] - 2026-09-19
+
+### Added
+
+- **Git-style context log** (`--context-log`, `src/core/agent/context-log.ts`, config `contextLog`, env `MMA_CONTEXT_LOG`): writes the full LLM context (system prompt + tool definitions + message history) to `<session>/context.diff`. Snapshot 0 is the complete context (every line `+`, like a new git file); each later snapshot is a plain unified line diff (`-` removed, `+` added) against the previous iteration. Sections are `[SYSTEM]`, `[TOOLS]`, `[HISTORY]`. Secrets are redacted via a dedicated, conservative pattern set (`redactContextSecrets`; `contextLog.redact`, default on) that targets recognizable tokens (sk-, ghp_, AKIA, JWT, Bearer, `api_key=`/`token=`/`password=`) and leaves ordinary context text (skill names, `pass "background: true"`, etc.) intact. The file is append-only: a second run adds a `run <pid> · <timestamp>` separator and restarts with a full snapshot while snapshot numbering continues. Model **reasoning/text/tool calls** are not part of the outgoing context and are recorded as separate `[OUTPUT]` annotations (`▍`), never diffed. Hooked in `Agent.executeSingleAgentLoop` next to the prefix monitor; no-op when disabled.
+
 ## [1.1.0] - 2026-09-19
 
 ### Added

package/dist/i18n/en.json

@@ -11,6 +11,10 @@
   "file.is_directory": "Is a directory, use rmdir: {path}",
   "file.replaced_in": "Replaced in {path}",
   "file.string_not_found": "String not found in file:\n{str}\n\nUse read_file on {path} to see the current content before editing — the target may differ (whitespace, line endings) or was already replaced.",
+  "file.edit_args_invalid": "edit_file requires string \"old\" and \"new\" arguments. Received: old={oldType}, new={newType}.",
+  "file.read_failed": "Could not read {path}: {error}",
+  "file.write_failed": "Could not write {path}: {error}",
+  "file.protected_path_ask": "Protected path \"{path}\" (matches {pattern}). Do NOT modify it. Ask the user for explicit confirmation before proceeding and wait for their answer — do not retry around this.",
   "file.moved": "Moved {from} → {to}",
   "file.not_found_short": "Not found: {path}",
   "file.notfound_resolved": "File not found: {path} (resolved to {resolved})\nThe path was joined onto the working directory because it does not exist as given. Run list_dir to find the correct path.",
@@ -47,6 +51,8 @@
   "error.llm_output_truncated": "The response hit the completion token limit ({tokens}) and was cut off — this answer is incomplete. Raise \"maxCompletionTokens\" in the config to allow longer replies",
   "error.llm_provider_stream_error": "Provider returned an error mid-stream: {error}",
   "error.llm_timeout": "LLM request timed out ({timeout}ms)",
+  "error.context_log_init": "Context log is enabled but could not be initialized: {error}. No context diff will be written.",
+  "error.provider_build": "Could not build provider \"{name}\": {error}",
   "env.runtime_node": "Running under Node (v{version}) — clipboard image paste, subagent performance and LSP spawn on Windows degrade. Install Bun (https://bun.sh) for full features.",
   "env.runtime_old": "Runtime version {version} is below the required engines {engine}.",
   "env.tool_missing": "Tool not found on PATH: {tool}",
@@ -92,6 +98,8 @@
   "tool.failed": "Tool {name} failed: {error}",
   "tool.unknown": "Unknown tool: {name}",
   "tool.blocked": "Blocked by plugin: {plugin}",
+  "tool.ui.security_blocked": "Blocked by security policy",
+  "tool.ui.protected_path": "Protected path — confirmation needed",
   "tool.blocked_reason": "Blocked by plugin: {plugin}. Reason: {reason}",
   "tool.using": "[{label}]",
   "tool.friendly.write_file": "Writing file",
@@ -351,6 +359,8 @@
   "cli.exit_on_complete": "Exit immediately when the task is done; interactive tools (question/approve) fail fast instead of blocking on stdin",
   "cli.reasoning_level": "Set reasoning effort level (auto|none|low|medium|high|max)",
   "cli.json": "Output the result as JSON (single-run mode)",
+  "cli.context_log": "Write the full LLM context to <session>/context.diff as a git-style diff (first snapshot full, then +/- changes)",
+  "cli.verbose": "Show full chat detail (tool bodies, lists); default is quiet",
   "cli.security.description": "Manage security settings",
   "cli.security.status": "Show current security configuration",
   "cli.security.policies": "List available security policies",
@@ -411,6 +421,11 @@
   "repl.reasoning_usage": "Usage: /reasoning [auto|none|low|medium|high|max]",
   "repl.reasoning_level_set": "Reasoning level set to: {level}",
   "repl.reasoning_invalid_level": "Invalid level '{level}'. Valid: {valid}",
+  "repl.verbose": "Chat verbosity: quiet, normal, or verbose",
+  "repl.verbose_usage": "Usage: /verbose [quiet|normal|verbose]",
+  "repl.verbose_set": "Chat verbosity: {level}",
+  "repl.verbose_invalid_level": "Invalid level '{level}'. Valid: {valid}",
+  "repl.last_action_none": "No recent action to show.",
   "repl.status": "Show agent status",
   "repl.status_usage": "Usage: /status",
   "repl.plugins": "List loaded plugins",
@@ -580,6 +595,9 @@
   "exec.audit_incomplete": "❗ Final audit incomplete: {summary}. Task is NOT finished. Remaining steps: {steps}. Resolve them - pick ONE path: (a) if the step's deliverable already exists on disk, mark it done (plan update step=N status=done); (b) if the step is no longer needed, mark it skipped; (c) if the plan drifted from reality (wrong file names, changed approach), re-plan to align it. Do NOT retry plan update repeatedly without changing anything.",
   "exec.mass_edit_warning": "⚠️ Plan affects {count} files — review the full list before proceeding.",
   "exec.escalation": "\n\n⚠️ Agent stuck on step {stepId} ({description}). Escalating to user — please provide guidance.",
+  "exec.escalation_directive": "You have been stuck for several attempts. Stop repeating the same action and pick ONE of these: (1) switch to a fundamentally different approach, (2) search the exact error text with web_search (or the error-search tool), (3) ask the user a short clarifying question, or (4) if the step cannot be completed, use plan update step=<n> status=skipped with a note (or re-plan). Explain your choice briefly.",
+  "exec.test_churn_nudge": "You have edited the test/spec file \"{file}\" several times in a row without changing any production code. If the tests fail, fix the PRODUCTION code under test (not the test) unless the test itself is genuinely wrong — and if so, explain why. If you intended test-first, now write the implementation.",
+  "exec.script_workaround_nudge": "A security restriction already blocked one of your commands. Writing a shell script (.bat/.cmd/.sh/.ps1) to run it another way does NOT lift the restriction and usually just fails. Instead: use write_file for file output, run one plain command per call, pass \"workdir\" instead of \"cd\", or ask the user.",
   "exec.hints": "\n[Hints]\n{hints}",
   "exec.file_rewrite_warning": "⚠️ File {file} has been rewritten {count} times. Consider a different approach — the current fix strategy is not working.",
   "exec.forbidden_cmd": "STOP using \"{cmd}\" via the bash tool — it is not a native Windows cmd.exe command and has failed repeatedly this session. Use the dedicated tool instead: grep → the grep tool, ls/dir → list_dir, find → glob, rm → delete_file, sed → edit_file, touch → write_file, which → `where`, cp/mv → move_file, diff → read_file. Do NOT call bash for this purpose again.",
@@ -632,10 +650,18 @@
   "migration.dir_bak": "- .mma/ → .mma.bak/ ({count} files)",
   "migration.migrated": "Migrated:",
   "migration.not_needed": "No migration needed",
-  "ui.error_prefix": "Error: ",
+  "ui.error_prefix": "✗ ",
   "ui.success_prefix": "✓ ",
   "ui.warning_prefix": "⚠ ",
   "ui.thinking": "Thinking…",
+  "ui.thinking_label": "thinking",
+  "ui.thinking_label_1": "thinking",
+  "ui.thinking_label_2": "working it out",
+  "ui.thinking_label_3": "considering",
+  "ui.action_fallback": "Action",
+  "ui.actions_summary": "{count} actions",
+  "ui.plan_update": "step {step} → {status}",
+  "ui.todo_action": "{action} · {count}",
   "ui.tool_running": "{tool}…",
   "ui.step_context": "step {id}: {desc}",
   "indexer.map_header": "Project map",
@@ -738,6 +764,7 @@
   "cli.usage_account": "Account: {credits} purchased · {used} used",
   "cli.usage_empty": "No balance information returned.",
   "repl.ctrl_c_interrupt": "\n[Ctrl+C] Stopping agent... (press again to force)",
+  "repl.force_exit": "\nForcing exit — background cleanup may be incomplete.",
   "exec.stop_directive": "STOP. Step {stepId} (\"{description}\") took {iterations} iterations with no progress. DO NOT continue this step. Immediately call: plan update step={stepId} status=done (if code works despite warnings) OR plan update step={stepId} status=skipped note=\"reason\". Do NOT make any other tool calls before updating the plan.",
   "exec.test_runner_fail": "[test-runner] {framework}: {failed} test(s) FAILING, {passed} passing — do NOT mark verification steps as done while tests fail. Investigate the failures, fix the code, then re-run the tests.",
   "exec.test_runner_pass": "[test-runner] {framework}: all {passed} test(s) passing.",
@@ -756,6 +783,11 @@
   "moe.execution_complete": "Execution complete: {succeeded}/{total} succeeded",
   "moe.cycles_exhausted": "Max re-plan cycles ({max}) reached — returning partial results",
   "moe.partial_result": "Partial result",
+  "moe.planning": "🤖 Planning with MoE mode...\n",
+  "moe.plan_fallback": "⚠️ Plan validation failed. Falling back to single-agent mode.\n",
+  "moe.plan_autofixed": "🔧 Auto-fixed {count} plan issues.\n",
+  "moe.plan_failed": "MoE plan failed: {error} — falling back to single-agent",
+  "moe.no_orchestrator": "MoE enabled but no orchestrator model configured — falling back to single-agent. Set orchestrator.model to activate MoE.",
   "moe.scope_approved": "🔓 Scope expanded for \"{subtask}\": {files}",
   "moe.scope_rejected": "⛔ Scope request rejected for \"{subtask}\"",
   "prompt.overflow.compressing": "System prompt overflow: compressing {count} block(s) before the first run — this may take a while on a local model.",

package/dist/i18n/ru.json

@@ -11,6 +11,10 @@
   "file.is_directory": "Это каталог, используйте rmdir: {path}",
   "file.replaced_in": "Заменено в {path}",
   "file.string_not_found": "Строка не найдена в файле:\n{str}\n\nИспользуй read_file для {path}, чтобы увидеть текущее содержимое файла перед редактированием — искомая строка может отличаться (пробелы, переводы строк) или уже была заменена.",
+  "file.edit_args_invalid": "edit_file требует строковые аргументы \"old\" и \"new\". Получено: old={oldType}, new={newType}.",
+  "file.read_failed": "Не удалось прочитать {path}: {error}",
+  "file.write_failed": "Не удалось записать {path}: {error}",
+  "file.protected_path_ask": "Защищённый путь \"{path}\" (совпадает с {pattern}). НЕ изменяй его. Спроси у пользователя явного подтверждения и дождись ответа — не обходи это ограничение.",
   "file.moved": "Перемещён {from} → {to}",
   "file.not_found_short": "Не найдено: {path}",
   "file.empty": "(пусто)",
@@ -46,6 +50,8 @@
   "error.llm_output_truncated": "Ответ упёрся в лимит токенов генерации ({tokens}) и был обрезан — этот ответ неполный. Увеличьте \"maxCompletionTokens\" в конфиге, чтобы разрешить длинные ответы",
   "error.llm_provider_stream_error": "Провайдер вернул ошибку посреди стрима: {error}",
   "error.llm_timeout": "Время запроса LLM истекло ({timeout}мс)",
+  "error.context_log_init": "Context log включён, но не удалось инициализировать: {error}. Diff контекста писаться не будет.",
+  "error.provider_build": "Не удалось собрать провайдера \"{name}\": {error}",
   "env.runtime_node": "Запущено под Node (v{version}) — вставка изображений из буфера и LSP на Windows работают урезанно. Установите Bun (https://bun.sh) для полного функционала.",
   "env.runtime_old": "Версия рантайма {version} ниже требуемой engines {engine}.",
   "env.tool_missing": "Инструмент не найден в PATH: {tool}",
@@ -91,6 +97,8 @@
   "tool.failed": "Инструмент {name} упал: {error}",
   "tool.unknown": "Неизвестный инструмент: {name}",
   "tool.blocked": "Заблокировано плагином: {plugin}",
+  "tool.ui.security_blocked": "Заблокировано политикой безопасности",
+  "tool.ui.protected_path": "Защищённый путь — нужно подтверждение",
   "tool.blocked_reason": "Заблокировано плагином: {plugin}. Причина: {reason}",
   "tool.using": "[{label}]",
   "tool.friendly.write_file": "Запись файла",
@@ -345,6 +353,8 @@
   "cli.exit_on_complete": "Завершить сразу после выполнения задачи; интерактивные тулзы (question/approve) не блокируют stdin, а возвращают ошибку",
   "cli.reasoning_level": "Установить уровень рассуждений (auto|none|low|medium|high|max)",
   "cli.json": "Вывести результат в формате JSON (режим однократного запуска)",
+  "cli.context_log": "Писать весь контекст LLM в <session>/context.diff в git-подобном diff-формате (первый снимок — полный, далее изменения +/-)",
+  "cli.verbose": "Показывать полный вывод чата (тела тулов, списки); по умолчанию — кратко",
   "cli.security.description": "Управление настройками безопасности",
   "cli.security.status": "Показать текущую конфигурацию безопасности",
   "cli.security.policies": "Список доступных политик безопасности",
@@ -405,6 +415,11 @@
   "repl.reasoning_usage": "Использование: /reasoning [auto|none|low|medium|high|max]",
   "repl.reasoning_level_set": "Уровень рассуждений установлен: {level}",
   "repl.reasoning_invalid_level": "Неверный уровень '{level}'. Допустимые: {valid}",
+  "repl.verbose": "Подробность вывода: quiet, normal или verbose",
+  "repl.verbose_usage": "Использование: /verbose [quiet|normal|verbose]",
+  "repl.verbose_set": "Подробность вывода: {level}",
+  "repl.verbose_invalid_level": "Неверный уровень '{level}'. Допустимые: {valid}",
+  "repl.last_action_none": "Нет недавнего действия для показа.",
   "repl.status": "Показать статус агента",
   "repl.status_usage": "Использование: /status",
   "repl.plugins": "Список загруженных плагинов",
@@ -578,6 +593,9 @@
   "exec.audit_incomplete": "❗ Финальная проверка не пройдена: {summary}. Задача НЕ завершена. Незакрытые шаги: {steps}. Устрани одним путём: (a) если результат шага уже существует на диске — закрой шаг (plan update step=N status=done); (b) если шаг больше не нужен — status=skipped; (c) если план разошёлся с реальностью (другие имена файлов, сменился подход) — сделай re-plan. Не повторяй plan update без изменений.",
   "exec.mass_edit_warning": "⚠️ План затрагивает {count} файлов — проверьте полный список перед продолжением.",
   "exec.escalation": "\n\n⚠️ Агент застрял на шаге {stepId} ({description}). Эскалация к пользователю — пожалуйста, подскажите как действовать.",
+  "exec.escalation_directive": "Ты застрял уже несколько попыток. Перестань повторять одно и то же и выбери ОДНО: (1) смени подход принципиально, (2) найди точный текст ошибки через web_search (или инструмент поиска ошибок), (3) задай пользователю короткий уточняющий вопрос, (4) если шаг невыполним — plan update step=<n> status=skipped с заметкой (или re-plan). Кратко объясни выбор.",
+  "exec.test_churn_nudge": "Ты несколько раз подряд правил тест/спек-файл \"{file}\" и не изменил ни строчки production-кода. Если тесты падают — исправляй КОД ПОД ТЕСТОМ (а не тест), если только сам тест не ошибочен — тогда объясни, почему. Если задумано test-first, теперь напиши реализацию.",
+  "exec.script_workaround_nudge": "Ограничение безопасности уже заблокировало одну из твоих команд. Написание shell-скрипта (.bat/.cmd/.sh/.ps1), чтобы выполнить её иначе, НЕ снимает ограничение и обычно просто падает. Вместо этого: используй write_file для вывода в файл, запускай по одной простой команде за вызов, передавай \"workdir\" вместо \"cd\" или спроси пользователя.",
   "exec.hints": "\n[Подсказки]\n{hints}",
   "exec.file_rewrite_warning": "⚠️ Файл {file} был перезаписан {count} раз. Попробуйте другой подход — текущая стратегия исправлений не работает.",
   "exec.forbidden_cmd": "ПРЕКРАТИ использовать \"{cmd}\" через bash — это не команда Windows cmd.exe, и она уже неоднократно падала в этой сессии. Используй предназначенный тул: grep → тул grep, ls/dir → list_dir, find → glob, rm → delete_file, sed → edit_file, touch → write_file, which → `where`, cp/mv → move_file, diff → read_file. Больше не вызывай bash для этого.",
@@ -630,10 +648,18 @@
   "migration.dir_bak": "- .mma/ → .mma.bak/ ({count} файлов)",
   "migration.migrated": "Мигрировано:",
   "migration.not_needed": "Миграция не требуется",
-  "ui.error_prefix": "Ошибка: ",
+  "ui.error_prefix": "✗ ",
   "ui.success_prefix": "✓ ",
   "ui.warning_prefix": "⚠ ",
   "ui.thinking": "Думаю…",
+  "ui.thinking_label": "размышляю",
+  "ui.thinking_label_1": "размышляю",
+  "ui.thinking_label_2": "обдумываю",
+  "ui.thinking_label_3": "прикидываю",
+  "ui.action_fallback": "Действие",
+  "ui.actions_summary": "{count} действий",
+  "ui.plan_update": "шаг {step} → {status}",
+  "ui.todo_action": "{action} · {count}",
   "ui.tool_running": "{tool}…",
   "ui.step_context": "шаг {id}: {desc}",
   "indexer.map_header": "Карта проекта",
@@ -738,6 +764,7 @@
   "cli.usage_account": "Аккаунт: куплено {credits} · израсходовано {used}",
   "cli.usage_empty": "Провайдер не вернул данных о балансе.",
   "repl.ctrl_c_interrupt": "\n[Ctrl+C] Остановка агента... (ещё раз — принудительно)",
+  "repl.force_exit": "\nПринудительный выход — фоновая очистка может быть неполной.",
   "exec.stop_directive": "STOP. Step {stepId} (\"{description}\") took {iterations} iterations with no progress. DO NOT continue this step. Immediately call: plan update step={stepId} status=done (if code works despite warnings) OR plan update step={stepId} status=skipped note=\"reason\". Do NOT make any other tool calls before updating the plan.",
   "exec.test_runner_fail": "[test-runner] {framework}: {failed} test(s) FAILING, {passed} passing — do NOT mark verification steps as done while tests fail. Investigate the failures, fix the code, then re-run the tests.",
   "exec.test_runner_pass": "[test-runner] {framework}: all {passed} test(s) passing.",
@@ -756,6 +783,11 @@
   "moe.execution_complete": "Выполнение завершено: {succeeded}/{total} успешно",
   "moe.cycles_exhausted": "Достигнут лимит re-plan циклов ({max}) — возвращаю частичные результаты",
   "moe.partial_result": "Частичный результат",
+  "moe.planning": "🤖 Планирую в режиме MoE...\n",
+  "moe.plan_fallback": "⚠️ Валидация плана не прошла. Возврат к одиночному режиму.\n",
+  "moe.plan_autofixed": "🔧 Автоисправлено проблем плана: {count}.\n",
+  "moe.plan_failed": "План MoE не построен: {error} — возврат к одиночному режиму",
+  "moe.no_orchestrator": "MoE включён, но orchestrator.model не задан — возврат к одиночному режиму. Укажите orchestrator.model для активации MoE.",
   "moe.scope_approved": "🔓 Scope расширен для \"{subtask}\": {files}",
   "moe.scope_rejected": "⛔ Запрос scope отклонён для \"{subtask}\"",
   "prompt.overflow.compressing": "Переполнение системного промпта: сжимаю {count} блок(ов) перед первым запуском — на локальной модели это может занять время.",