@blade-ai/orca 0.1.3-linux-x64 → 0.1.6-darwin-arm64

package/README.md

@@ -1,6 +1,6 @@
 # Orca
 
-Orca is a DeepSeek-native coding agent runtime by Blade.
+Orca is a DeepSeek-native coding agent.
 
 A local terminal coding agent built in Rust, focused on DeepSeek reasoning and tool-use semantics. It runs a multi-turn agent loop with SSE streaming, automatic context window management, and HTTP retry with exponential backoff.
 
@@ -25,15 +25,15 @@ Supported npm platforms:
 ### curl
 
 ```bash
-curl -fsSL https://raw.githubusercontent.com/echoVic/blade-deepseek/main/install.sh | sh
+curl -fsSL https://orcaagent.dev/install.sh | sh
 ```
 
 The installer downloads the native binary for your platform from GitHub Releases.
 Set `INSTALL_DIR` to choose a destination and `ORCA_VERSION` to pin a version:
 
 ```bash
-curl -fsSL https://raw.githubusercontent.com/echoVic/blade-deepseek/main/install.sh | \
-  INSTALL_DIR=/usr/local/bin ORCA_VERSION=0.1.3 sh
+curl -fsSL https://orcaagent.dev/install.sh | \
+  INSTALL_DIR=/usr/local/bin ORCA_VERSION=0.1.6 sh
 ```
 
 ### GitHub Releases
@@ -159,6 +159,23 @@ orca --session-picker
 
 In the TUI, `Esc` during an idle composer backtracks to the previous user message and places that prompt back in the input box for editing and re-asking.
 
+## Persistent Goal Mode
+
+TUI sessions support Codex-style persistent goals with `/goal`. A goal is stored by session id in `~/.orca/goals_1.json` or `$ORCA_HOME/goals_1.json`, so it survives process restarts when the session is saved.
+
+```text
+/goal                         # show the current goal
+/goal ship the refactor       # create or replace the active goal and start it
+/goal edit finish the parser  # update the objective and reactivate it
+/goal pause                   # stop automatic continuation
+/goal resume                  # reactivate and continue when idle
+/goal clear                   # delete the goal for this session
+```
+
+While a goal is active, Orca automatically starts another turn after a successful turn and injects goal-mode instructions as pinned context. The loop stops when the goal is paused, cleared, blocked, completed, budget-limited, interrupted, or reaches the continuation cap. The model can stop the loop with the `update_goal` tool by setting status `complete` or `blocked`.
+
+Persistent goals require recorded history. If history is disabled with `--no-history`, `/goal` reports an error instead of creating ephemeral goal state.
+
 ## Tools
 
 Built-in tools:
@@ -170,15 +187,21 @@ Built-in tools:
 | `grep` | Search with ripgrep (regex, line numbers) |
 | `bash` | Execute shell commands via `sh -c` |
 | `edit` | Exact text replacement in files |
+| `write_file` | Create or overwrite a file |
 | `git_status` | Show git working tree status |
+| `web_search` | Search the web for current information |
 | `subagent` | Run a synchronous child agent for a delegated task |
+| `Workflow` | Launch a background dynamic workflow |
+| `update_plan` | Update the visible task plan |
+| `update_goal` | Update active persistent goal status from goal mode |
 
 ## Architecture
 
 - **Agent Loop**: prompt → model → tool_call → execute → feed result → next turn (up to 128 turns)
 - **Subagents**: Synchronous child agent loops share the parent workspace, provider/model config, and approval policy, then return a concise result to the parent
+- **Persistent Goal Mode**: TUI sessions can persist a long-running objective, auto-continue successful turns, and stop through `/goal` controls or the `update_goal` tool
 - **SSE Streaming**: Real-time reasoning and content deltas via Server-Sent Events
-- **Context Window**: 128K tokens, 80% threshold compaction (preserves system + recent messages)
+- **Context Window**: DeepSeek V4 1M-token context, 80% threshold compaction with response reserve (preserves system + recent messages)
 - **Conversation History**: Local JSONL transcripts support listing, inspection, resume/fork, full-text search, archive/delete/rename, and zstd compression
 - **HTTP Client**: Singleton with 30s connect / 120s request / 300s streaming timeouts, exponential backoff retry (3 attempts, handles 429/5xx)
 - **Approval Policy**: Read operations always allowed; write/shell actions require interactive confirmation (suggest mode) or auto-allowed based on mode
@@ -211,4 +234,4 @@ Event types: `session.started`, `turn.started`, `assistant.reasoning.delta`, `as
 
 ## Status
 
-Production-ready agent loop with DeepSeek streaming provider. All 7 tools implemented, multi-turn conversation with context management, subagents, approval policies, and verification support.
+Production-ready agent loop with DeepSeek streaming provider. Core tools, multi-turn conversation, persistent goals, context management, subagents, workflows, approval policies, and verification support are implemented.

package/package.json

@@ -12,12 +12,12 @@
     "node": ">=16"
   },
   "name": "@blade-ai/orca",
-  "version": "0.1.3-linux-x64",
-  "description": "Native Orca binary for linux/x64.",
+  "version": "0.1.6-darwin-arm64",
+  "description": "Native Orca binary for darwin/arm64.",
   "os": [
-    "linux"
+    "darwin"
   ],
   "cpu": [
-    "x64"
+    "arm64"
   ]
 }