akernel-runtime 0.1.8__tar.gz → 0.1.9__tar.gz

akernel_runtime-0.1.9/.github/release-notes/v0.1.9.md

@@ -0,0 +1,14 @@
+# Context Kernel v0.1.9
+
+This release polishes the default `akernel` chat experience and adds more Claude/Gemini-style terminal agent workflows.
+
+## Highlights
+
+- Reworked the default TUI startup into a quieter chat-first header while keeping the full cockpit dashboard available with `AKERNEL_ALT_SCREEN=1`.
+- Added inline `@path` attachment inside normal chat tasks, so exact workspace files can be referenced without a separate attach command.
+- Added project and user Markdown slash commands under `.akernel/commands` and `~/.akernel/commands`, with `/commands` discovery and completion.
+- Improved `/` and `@` completion so suggestions work on the current cursor token, including inline text such as `inspect @README`.
+
+## Validation
+
+- `python -m unittest discover -s tests`

{akernel_runtime-0.1.8 → akernel_runtime-0.1.9}/CHANGELOG.md

@@ -8,6 +8,18 @@ The project follows a pragmatic pre-1.0 changelog: breaking changes may occur, b
 
 No changes yet.
 
+## 0.1.9 - 2026-05-14
+
+### Added
+
+- Added inline `@path` attachment inside normal chat tasks, so exact workspace files can be referenced without a separate attach command.
+- Added project and user Markdown slash commands under `.akernel/commands` and `~/.akernel/commands`, with `/commands` discovery and completion.
+
+### Changed
+
+- Reworked the default TUI startup into a quieter chat-first header while keeping the full cockpit dashboard available with `AKERNEL_ALT_SCREEN=1`.
+- Improved command and file completion so `/` and `@` work on the current cursor token, including inline text such as `inspect @README`.
+
 ## 0.1.8 - 2026-05-14
 
 ### Changed

{akernel_runtime-0.1.8/src/akernel_runtime.egg-info → akernel_runtime-0.1.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: akernel-runtime
-Version: 0.1.8
+Version: 0.1.9
 Summary: Agent Kernel: a CLI-first context-native agent runtime prototype.
 Author: Context Kernel contributors
 License-Expression: Apache-2.0
@@ -135,7 +135,7 @@ akernel
 
 ### npm Launcher
 
-The npm launcher is prepared for publication. After `@context-akernel/akernel` is published, it will provide a Node-style entrypoint and bootstrap the Python package if it is missing:
+The npm launcher is published as `@context-akernel/akernel`. It provides a Node-style entrypoint and bootstraps the Python package if it is missing:
 
 ```powershell
 npm install -g @context-akernel/akernel
@@ -155,7 +155,9 @@ akernel --ui tui
 akernel --provider mock
 ```
 
-Inside the interactive session, type a task and press Enter. Bare `akernel` accepts chat flags such as `--provider mock`, `--model`, `--aux-model`, `--max-steps`, and `--ui tui` without requiring the explicit `chat` subcommand. `--ui auto` uses the full-screen terminal UI on real terminals and falls back to classic output for CI, pipes, and tests. `akernel init . --scan` or `akernel project scan` writes a compact `.akernel/project.json` profile with detected languages, package managers, key files, project instruction files such as `AGENTS.md`, and test/build commands; this profile enters future context packets without loading the whole repository. If you ask to `run tests` or `verify` without naming a command, the agent prefers the scanned project test command. If you ask it to fix failing tests, the agent can run that profile test command, inspect one or more failing files from the command output, apply a bounded patch or rollback-safe batch patch when the failure is simple enough, and rerun verification. Use `/status` for the live workspace view, `/model` for primary and auxiliary model roles, `/cost` for the last run's token report, `/task` to inspect the current task session, and `/exit` to leave. The shell UI renders a cockpit-style dashboard with a status header, command strip, transcript, mission panel, model stack, workspace summary, task progress, run timeline, diagnostics, and assistant responses. In `--model-routing auto` mode, low/medium first-step planning can run on the auxiliary model while high-risk, deep, warning-heavy, or synthesis steps stay on the primary model. In `--aux-review auto` mode, auxiliary review runs before primary-model steps and is included in token cost reports.
+Inside the interactive session, type a task and press Enter. Bare `akernel` accepts chat flags such as `--provider mock`, `--model`, `--aux-model`, `--max-steps`, and `--ui tui` without requiring the explicit `chat` subcommand. `--ui auto` uses the polished terminal chat UI on real terminals and falls back to classic output for CI, pipes, and tests. `akernel init . --scan` or `akernel project scan` writes a compact `.akernel/project.json` profile with detected languages, package managers, key files, project instruction files such as `AGENTS.md`, and test/build commands; this profile enters future context packets without loading the whole repository. If you ask to `run tests` or `verify` without naming a command, the agent prefers the scanned project test command. If you ask it to fix failing tests, the agent can run that profile test command, inspect one or more failing files from the command output, apply a bounded patch or rollback-safe batch patch when the failure is simple enough, and rerun verification.
+
+The default chat UI is scrollback-friendly: it shows a concise session header once, then appends only the user turn, compact run status, and assistant result. Type `/status` for the live workspace view, `/model` for primary and auxiliary model roles, `/cost` for the last run's token report, `/task` to inspect the current task session, and `/exit` to leave. Type `/` for command completion, `@` for workspace file completion, or mention an exact `@path` inside a task to attach that file automatically. Project and user slash commands can be saved as Markdown prompts under `.akernel/commands` or `~/.akernel/commands`; use `/commands` to list them. Set `AKERNEL_ALT_SCREEN=1` if you prefer the older full cockpit dashboard with a side panel, transcript viewport, mission panel, model stack, workspace summary, task progress, run timeline, diagnostics, and assistant responses. In `--model-routing auto` mode, low/medium first-step planning can run on the auxiliary model while high-risk, deep, warning-heavy, or synthesis steps stay on the primary model. In `--aux-review auto` mode, auxiliary review runs before primary-model steps and is included in token cost reports.
 
 If you want to prepare the benchmark workspace manually:
 

{akernel_runtime-0.1.8 → akernel_runtime-0.1.9}/README.md

@@ -107,7 +107,7 @@ akernel
 
 ### npm Launcher
 
-The npm launcher is prepared for publication. After `@context-akernel/akernel` is published, it will provide a Node-style entrypoint and bootstrap the Python package if it is missing:
+The npm launcher is published as `@context-akernel/akernel`. It provides a Node-style entrypoint and bootstraps the Python package if it is missing:
 
 ```powershell
 npm install -g @context-akernel/akernel
@@ -127,7 +127,9 @@ akernel --ui tui
 akernel --provider mock
 ```
 
-Inside the interactive session, type a task and press Enter. Bare `akernel` accepts chat flags such as `--provider mock`, `--model`, `--aux-model`, `--max-steps`, and `--ui tui` without requiring the explicit `chat` subcommand. `--ui auto` uses the full-screen terminal UI on real terminals and falls back to classic output for CI, pipes, and tests. `akernel init . --scan` or `akernel project scan` writes a compact `.akernel/project.json` profile with detected languages, package managers, key files, project instruction files such as `AGENTS.md`, and test/build commands; this profile enters future context packets without loading the whole repository. If you ask to `run tests` or `verify` without naming a command, the agent prefers the scanned project test command. If you ask it to fix failing tests, the agent can run that profile test command, inspect one or more failing files from the command output, apply a bounded patch or rollback-safe batch patch when the failure is simple enough, and rerun verification. Use `/status` for the live workspace view, `/model` for primary and auxiliary model roles, `/cost` for the last run's token report, `/task` to inspect the current task session, and `/exit` to leave. The shell UI renders a cockpit-style dashboard with a status header, command strip, transcript, mission panel, model stack, workspace summary, task progress, run timeline, diagnostics, and assistant responses. In `--model-routing auto` mode, low/medium first-step planning can run on the auxiliary model while high-risk, deep, warning-heavy, or synthesis steps stay on the primary model. In `--aux-review auto` mode, auxiliary review runs before primary-model steps and is included in token cost reports.
+Inside the interactive session, type a task and press Enter. Bare `akernel` accepts chat flags such as `--provider mock`, `--model`, `--aux-model`, `--max-steps`, and `--ui tui` without requiring the explicit `chat` subcommand. `--ui auto` uses the polished terminal chat UI on real terminals and falls back to classic output for CI, pipes, and tests. `akernel init . --scan` or `akernel project scan` writes a compact `.akernel/project.json` profile with detected languages, package managers, key files, project instruction files such as `AGENTS.md`, and test/build commands; this profile enters future context packets without loading the whole repository. If you ask to `run tests` or `verify` without naming a command, the agent prefers the scanned project test command. If you ask it to fix failing tests, the agent can run that profile test command, inspect one or more failing files from the command output, apply a bounded patch or rollback-safe batch patch when the failure is simple enough, and rerun verification.
+
+The default chat UI is scrollback-friendly: it shows a concise session header once, then appends only the user turn, compact run status, and assistant result. Type `/status` for the live workspace view, `/model` for primary and auxiliary model roles, `/cost` for the last run's token report, `/task` to inspect the current task session, and `/exit` to leave. Type `/` for command completion, `@` for workspace file completion, or mention an exact `@path` inside a task to attach that file automatically. Project and user slash commands can be saved as Markdown prompts under `.akernel/commands` or `~/.akernel/commands`; use `/commands` to list them. Set `AKERNEL_ALT_SCREEN=1` if you prefer the older full cockpit dashboard with a side panel, transcript viewport, mission panel, model stack, workspace summary, task progress, run timeline, diagnostics, and assistant responses. In `--model-routing auto` mode, low/medium first-step planning can run on the auxiliary model while high-risk, deep, warning-heavy, or synthesis steps stay on the primary model. In `--aux-review auto` mode, auxiliary review runs before primary-model steps and is included in token cost reports.
 
 If you want to prepare the benchmark workspace manually:
 

{akernel_runtime-0.1.8 → akernel_runtime-0.1.9}/packages/npm/akernel/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@context-akernel/akernel",
-  "version": "0.1.8",
+  "version": "0.1.9",
   "description": "npm launcher wrapper for the Context Kernel akernel CLI.",
   "license": "Apache-2.0",
   "homepage": "https://github.com/huanxin0825-ctrl/context-akernel#readme",

{akernel_runtime-0.1.8 → akernel_runtime-0.1.9}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "akernel-runtime"
-version = "0.1.8"
+version = "0.1.9"
 description = "Agent Kernel: a CLI-first context-native agent runtime prototype."
 readme = "README.md"
 license = "Apache-2.0"

{akernel_runtime-0.1.8 → akernel_runtime-0.1.9/src/akernel_runtime.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: akernel-runtime
-Version: 0.1.8
+Version: 0.1.9
 Summary: Agent Kernel: a CLI-first context-native agent runtime prototype.
 Author: Context Kernel contributors
 License-Expression: Apache-2.0
@@ -135,7 +135,7 @@ akernel
 
 ### npm Launcher
 
-The npm launcher is prepared for publication. After `@context-akernel/akernel` is published, it will provide a Node-style entrypoint and bootstrap the Python package if it is missing:
+The npm launcher is published as `@context-akernel/akernel`. It provides a Node-style entrypoint and bootstraps the Python package if it is missing:
 
 ```powershell
 npm install -g @context-akernel/akernel
@@ -155,7 +155,9 @@ akernel --ui tui
 akernel --provider mock
 ```
 
-Inside the interactive session, type a task and press Enter. Bare `akernel` accepts chat flags such as `--provider mock`, `--model`, `--aux-model`, `--max-steps`, and `--ui tui` without requiring the explicit `chat` subcommand. `--ui auto` uses the full-screen terminal UI on real terminals and falls back to classic output for CI, pipes, and tests. `akernel init . --scan` or `akernel project scan` writes a compact `.akernel/project.json` profile with detected languages, package managers, key files, project instruction files such as `AGENTS.md`, and test/build commands; this profile enters future context packets without loading the whole repository. If you ask to `run tests` or `verify` without naming a command, the agent prefers the scanned project test command. If you ask it to fix failing tests, the agent can run that profile test command, inspect one or more failing files from the command output, apply a bounded patch or rollback-safe batch patch when the failure is simple enough, and rerun verification. Use `/status` for the live workspace view, `/model` for primary and auxiliary model roles, `/cost` for the last run's token report, `/task` to inspect the current task session, and `/exit` to leave. The shell UI renders a cockpit-style dashboard with a status header, command strip, transcript, mission panel, model stack, workspace summary, task progress, run timeline, diagnostics, and assistant responses. In `--model-routing auto` mode, low/medium first-step planning can run on the auxiliary model while high-risk, deep, warning-heavy, or synthesis steps stay on the primary model. In `--aux-review auto` mode, auxiliary review runs before primary-model steps and is included in token cost reports.
+Inside the interactive session, type a task and press Enter. Bare `akernel` accepts chat flags such as `--provider mock`, `--model`, `--aux-model`, `--max-steps`, and `--ui tui` without requiring the explicit `chat` subcommand. `--ui auto` uses the polished terminal chat UI on real terminals and falls back to classic output for CI, pipes, and tests. `akernel init . --scan` or `akernel project scan` writes a compact `.akernel/project.json` profile with detected languages, package managers, key files, project instruction files such as `AGENTS.md`, and test/build commands; this profile enters future context packets without loading the whole repository. If you ask to `run tests` or `verify` without naming a command, the agent prefers the scanned project test command. If you ask it to fix failing tests, the agent can run that profile test command, inspect one or more failing files from the command output, apply a bounded patch or rollback-safe batch patch when the failure is simple enough, and rerun verification.
+
+The default chat UI is scrollback-friendly: it shows a concise session header once, then appends only the user turn, compact run status, and assistant result. Type `/status` for the live workspace view, `/model` for primary and auxiliary model roles, `/cost` for the last run's token report, `/task` to inspect the current task session, and `/exit` to leave. Type `/` for command completion, `@` for workspace file completion, or mention an exact `@path` inside a task to attach that file automatically. Project and user slash commands can be saved as Markdown prompts under `.akernel/commands` or `~/.akernel/commands`; use `/commands` to list them. Set `AKERNEL_ALT_SCREEN=1` if you prefer the older full cockpit dashboard with a side panel, transcript viewport, mission panel, model stack, workspace summary, task progress, run timeline, diagnostics, and assistant responses. In `--model-routing auto` mode, low/medium first-step planning can run on the auxiliary model while high-risk, deep, warning-heavy, or synthesis steps stay on the primary model. In `--aux-review auto` mode, auxiliary review runs before primary-model steps and is included in token cost reports.
 
 If you want to prepare the benchmark workspace manually:
 

{akernel_runtime-0.1.8 → akernel_runtime-0.1.9}/src/akernel_runtime.egg-info/SOURCES.txt

@@ -24,6 +24,7 @@ wake.ps1
 .github/release-notes/v0.1.6.md
 .github/release-notes/v0.1.7.md
 .github/release-notes/v0.1.8.md
+.github/release-notes/v0.1.9.md
 .github/workflows/ci.yml
 .github/workflows/release.yml
 docs/00-vision.md

{akernel_runtime-0.1.8 → akernel_runtime-0.1.9}/src/context_kernel/__init__.py

@@ -1,3 +1,3 @@
 """Context Kernel CLI-first agent runtime prototype."""
 
-__version__ = "0.1.8"
+__version__ = "0.1.9"

{akernel_runtime-0.1.8 → akernel_runtime-0.1.9}/src/context_kernel/cli.py

@@ -7,6 +7,7 @@ import io
 import json
 import os
 from pathlib import Path
+import re
 import shutil
 import sys
 import unicodedata
@@ -49,6 +50,7 @@ CHAT_COMMANDS: list[tuple[str, str]] = [
     ("/model", "show primary and auxiliary model roles"),
     ("/config", "show setup and environment guidance"),
     ("/compact", "show compact task brief"),
+    ("/commands", "list project and user slash commands"),
     ("/paste", "enter a multi-line task"),
     ("/task", "print current task session JSON"),
     ("/runs", "list recent agent runs"),
@@ -59,6 +61,7 @@ CHAT_COMMANDS: list[tuple[str, str]] = [
     ("/clear", "clear transcript"),
     ("/exit", "leave interactive session"),
 ]
+INLINE_FILE_REF_RE = re.compile(r"(?<![\w@])@([^\s,;:]+)")
 OPENAI_ENV_KEYS = {
     "api_key": "AKERNEL_OPENAI_API_KEY",
     "base_url": "AKERNEL_OPENAI_BASE_URL",
@@ -1111,6 +1114,9 @@ def cmd_chat(args: argparse.Namespace) -> None:
         if lowered == "/compact":
             print_task_brief_panel(tasks, task_id)
             continue
+        if lowered == "/commands":
+            print_custom_commands_panel(workspace.root)
+            continue
         if lowered == "/paste":
             pasted = read_paste_block()
             if not pasted:
@@ -1149,25 +1155,14 @@ def cmd_chat(args: argparse.Namespace) -> None:
                 print(render_agent_cost_report(build_agent_cost_report(last_report)))
             continue
 
+        attach_inline_file_references(workspace, tasks, task_id, request, pending_context)
+        custom_prompt = expand_custom_chat_command(workspace.root, request)
+        if custom_prompt is not None:
+            request = custom_prompt
         request_for_agent = merge_pending_context(request, pending_context)
         pending_context.clear()
         print_chat_turn_start(request_for_agent, args)
-        last_report = AgentLoop(workspace).run(
-            request_for_agent,
-            provider_name=args.provider,
-            budget=args.budget,
-            profile=args.profile,
-            model=args.model,
-            aux_model=args.aux_model,
-            model_routing=args.model_routing,
-            aux_review=args.aux_review,
-            base_url=args.base_url,
-            task_id=task_id,
-            max_steps=args.max_steps,
-            remember=not args.no_remember,
-            allow_over_budget=args.allow_over_budget,
-            expect_json=args.expect_json,
-        )
+        last_report = run_chat_agent(workspace, task_id, args, request_for_agent)
         print_chat_report(last_report)
 
 
@@ -1286,30 +1281,27 @@ def run_chat_loop_tui(
                 render_chat_tui_update(workspace, task_id, args, transcript, last_report, pending_context, status="ready", state=state, clear=use_alt_screen)
                 continue
 
+            state["scroll_offset"] = 0
+            attach_notice = capture_chat_output(
+                lambda: attach_inline_file_references(workspace, tasks, task_id, request, pending_context)
+            )
+            if attach_notice:
+                transcript.append({"role": "system", "title": "Attached Context", "text": attach_notice})
+                render_chat_tui_message(transcript[-1])
+                state["rendered_count"] = len(transcript)
+            custom_prompt = expand_custom_chat_command(workspace.root, request)
+            if custom_prompt is not None:
+                request = custom_prompt
             request_for_agent = merge_pending_context(request, pending_context)
             pending_context.clear()
-            state["scroll_offset"] = 0
             transcript.append({"role": "user", "title": "You", "text": request_for_agent})
             render_chat_tui_message({"role": "user", "title": "You", "text": request_for_agent})
+            state["rendered_count"] = len(transcript)
             render_chat_tui_status("running", args, pending_context)
-            last_report = AgentLoop(workspace).run(
-                request_for_agent,
-                provider_name=args.provider,
-                budget=args.budget,
-                profile=args.profile,
-                model=args.model,
-                aux_model=args.aux_model,
-                model_routing=args.model_routing,
-                aux_review=args.aux_review,
-                base_url=args.base_url,
-                task_id=task_id,
-                max_steps=args.max_steps,
-                remember=not args.no_remember,
-                allow_over_budget=args.allow_over_budget,
-                expect_json=args.expect_json,
-            )
+            last_report = run_chat_agent(workspace, task_id, args, request_for_agent)
             transcript.append({"role": "assistant", "title": "Assistant", "text": format_tui_report(last_report)})
             render_chat_tui_message({"role": "assistant", "title": "Assistant", "text": format_tui_report(last_report)})
+            state["rendered_count"] = len(transcript)
             render_chat_tui_status("ready", args, pending_context, last_report=last_report)
     finally:
         if use_alt_screen:
@@ -1345,6 +1337,9 @@ def handle_tui_command(
     if lowered == "/compact":
         transcript.append({"role": "system", "title": "Compact Brief", "text": capture_chat_output(lambda: print_task_brief_panel(tasks, task_id))})
         return True
+    if lowered == "/commands":
+        transcript.append({"role": "system", "title": "Slash Commands", "text": capture_chat_output(lambda: print_custom_commands_panel(workspace.root))})
+        return True
     if lowered == "/status":
         transcript.append({"role": "system", "title": "Status", "text": capture_chat_output(lambda: print_status_panel(workspace, task_id, args))})
         return True
@@ -1371,22 +1366,7 @@ def handle_tui_command(
             transcript.append({"role": "user", "title": "Pasted Task", "text": pasted})
             request_for_agent = merge_pending_context(pasted, pending_context)
             pending_context.clear()
-            report = AgentLoop(workspace).run(
-                request_for_agent,
-                provider_name=args.provider,
-                budget=args.budget,
-                profile=args.profile,
-                model=args.model,
-                aux_model=args.aux_model,
-                model_routing=args.model_routing,
-                aux_review=args.aux_review,
-                base_url=args.base_url,
-                task_id=task_id,
-                max_steps=args.max_steps,
-                remember=not args.no_remember,
-                allow_over_budget=args.allow_over_budget,
-                expect_json=args.expect_json,
-            )
+            report = run_chat_agent(workspace, task_id, args, request_for_agent)
             state["last_report"] = report
             transcript.append({"role": "assistant", "title": "Assistant", "text": format_tui_report(report)})
         return True
@@ -1408,10 +1388,30 @@ def capture_chat_output(func: Any) -> str:
     return buffer.getvalue().strip()
 
 
+def run_chat_agent(workspace: Workspace, task_id: str, args: argparse.Namespace, request: str) -> dict[str, Any]:
+    return AgentLoop(workspace).run(
+        request,
+        provider_name=args.provider,
+        budget=args.budget,
+        profile=args.profile,
+        model=args.model,
+        aux_model=args.aux_model,
+        model_routing=args.model_routing,
+        aux_review=args.aux_review,
+        base_url=args.base_url,
+        task_id=task_id,
+        max_steps=args.max_steps,
+        remember=not args.no_remember,
+        allow_over_budget=args.allow_over_budget,
+        expect_json=args.expect_json,
+    )
+
+
 def format_chat_help_text() -> str:
     rows = [
         *CHAT_COMMANDS,
         ("@query", "search current workspace files; use @1, @2... to attach a listed match"),
+        ("@path", "mention an exact file path inside a task to attach it automatically"),
         ("!command", "run a policy-checked command and attach its summary"),
     ]
     return "\n".join(f"{name:<10} {description}" for name, description in rows)
@@ -1425,21 +1425,27 @@ def read_chat_input(prompt: str, workspace: Workspace) -> str:
         from prompt_toolkit.completion import Completer, Completion
         from prompt_toolkit.formatted_text import ANSI
         from prompt_toolkit.shortcuts import CompleteStyle
+        from prompt_toolkit.styles import Style
     except ImportError:
         return input(prompt)
 
     class ChatCompleter(Completer):
         def get_completions(self, document: Any, complete_event: Any) -> Any:
             text = document.text_before_cursor
+            fragment = chat_completion_fragment(text)
+            if not fragment:
+                return
             for value, description in chat_completion_items(workspace.root, text):
-                yield Completion(value, start_position=-len(text), display=value, display_meta=description)
+                yield Completion(value, start_position=-len(fragment), display=value, display_meta=description)
 
     session = PromptSession(
         completer=ChatCompleter(),
         complete_while_typing=True,
         complete_in_thread=True,
         complete_style=CompleteStyle.MULTI_COLUMN,
+        bottom_toolbar=" / opens commands   @ finds files   Tab accepts   Ctrl-C interrupts ",
         reserve_space_for_menu=6,
+        style=Style.from_dict({"bottom-toolbar": "reverse ansiblack ansibrightcyan"}),
     )
     return session.prompt(ANSI(prompt))
 
@@ -1450,15 +1456,26 @@ def should_use_prompt_toolkit() -> bool:
     return bool(sys.stdin.isatty() and sys.stdout.isatty())
 
 
+def chat_completion_fragment(text: str) -> str:
+    match = re.search(r"(^|\s)([/@][^\s]*)$", text)
+    return match.group(2) if match else ""
+
+
 def chat_completion_items(root: Path, text: str, *, limit: int = 12) -> list[tuple[str, str]]:
-    value = text.strip()
+    value = chat_completion_fragment(text) or text.strip()
     if value.startswith("/"):
         query = value.casefold()
-        return [
+        builtins = [
             (command, description)
             for command, description in CHAT_COMMANDS
             if command.casefold().startswith(query)
-        ][:limit]
+        ]
+        custom = [
+            (command, f"{spec['scope']} command: {spec['description']}")
+            for command, spec in load_custom_chat_commands(root).items()
+            if command.casefold().startswith(query)
+        ]
+        return (builtins + custom)[:limit]
     if value.startswith("@"):
         query = value[1:].strip()
         matches = find_workspace_files(root, query, limit=limit)
@@ -1485,6 +1502,8 @@ def render_chat_tui_screen(
     screen = build_chat_tui_screen(workspace, task_id, args, transcript, last_report, pending_context, status=status, state=state)
     prefix = "\033[2J\033[H" if clear else "\n"
     print(prefix + screen + ("\n" if not clear else ""), end="")
+    if state is not None:
+        state["rendered_count"] = len(transcript)
 
 
 def render_chat_tui_update(
@@ -1502,8 +1521,11 @@ def render_chat_tui_update(
     if clear:
         render_chat_tui_screen(workspace, task_id, args, transcript, last_report, pending_context, status=status, state=state, clear=True)
         return
-    if transcript:
-        render_chat_tui_message(transcript[-1])
+    start = int((state or {}).get("rendered_count", max(0, len(transcript) - 1)))
+    for item in transcript[start:]:
+        render_chat_tui_message(item)
+    if state is not None:
+        state["rendered_count"] = len(transcript)
     render_chat_tui_status(status, args, pending_context, last_report=last_report)
 
 
@@ -1543,9 +1565,9 @@ def build_chat_tui_screen(
     state: dict[str, Any] | None = None,
 ) -> str:
     width = chat_width()
-    height = max(24, shutil.get_terminal_size((width, 32)).lines)
     if (state or {}).get("scrollback_mode"):
-        height = min(height, 22)
+        return "\n".join(tui_compact_start_lines(workspace, task_id, args, last_report, pending_context, status=status, width=width))
+    height = max(24, shutil.get_terminal_size((width, 32)).lines)
     right_width = min(40, max(32, width // 3))
     left_width = max(46, width - right_width - 3)
     header = tui_header_lines(workspace, args, last_report, status=status, width=width)
@@ -1565,6 +1587,34 @@ def build_chat_tui_screen(
     return "\n".join(lines)
 
 
+def tui_compact_start_lines(
+    workspace: Workspace,
+    task_id: str,
+    args: argparse.Namespace,
+    last_report: dict[str, Any] | None,
+    pending_context: list[str],
+    *,
+    status: str,
+    width: int,
+) -> list[str]:
+    tokens = 0 if not last_report else last_report.get("totals", {}).get("total_tokens", 0)
+    task_short = task_id[:12]
+    return [
+        "",
+        chat_color(truncate_line("AKERNEL", width), "cyan", bold=True),
+        chat_color(truncate_line("Token-frugal agent workspace. Focused chat, explicit context, durable memory.", width), "dim"),
+        "",
+        truncate_line(f"cwd      {compact_path(Path.cwd())}", width),
+        truncate_line(f"work     {compact_path(workspace.root)}", width),
+        truncate_line(f"models   primary {primary_model(args)} | aux {auxiliary_model(args)} | routing {args.model_routing}", width),
+        truncate_line(f"session  task {task_short} | provider {args.provider} | profile {args.profile} | tokens {tokens} | status {status}", width),
+        truncate_line(f"context  attached {len(pending_context)} | type @ to search files, or mention @path inside a task", width),
+        "",
+        chat_color(truncate_line("shortcuts /help /status /model /commands /compact /runs /cost /clear /exit", width), "dim"),
+        chat_color(truncate_line("input     ask one concrete task; use !command for checked shell context", width), "dim"),
+    ]
+
+
 def tui_header_lines(
     workspace: Workspace,
     args: argparse.Namespace,
@@ -1858,8 +1908,10 @@ def print_chat_help() -> None:
             ("/model", "show primary and auxiliary model roles"),
             ("/config", "show setup and environment guidance"),
             ("/compact", "show the compact task brief used for resume context"),
+            ("/commands", "list saved project and user slash commands"),
             ("/paste", "enter a multi-line task; finish with /end"),
             ("@query", "search workspace files; use @1, @2... to attach a listed match"),
+            ("@path", "mention an exact file path inside a task to attach it automatically"),
             ("!command", "run a policy-checked command and attach its summary"),
             ("/task", "print the current task session JSON"),
             ("/runs", "list recent agent runs"),
@@ -1893,6 +1945,100 @@ def print_recent_agent_runs(workspace: Workspace, *, limit: int) -> None:
         )
 
 
+def load_custom_chat_commands(root: Path) -> dict[str, dict[str, str]]:
+    commands: dict[str, dict[str, str]] = {}
+    for directory, scope in custom_command_directories(root):
+        if not directory.exists():
+            continue
+        for path in sorted(directory.rglob("*.md"))[:80]:
+            try:
+                relative = path.relative_to(directory).with_suffix("").as_posix()
+                body = path.read_text(encoding="utf-8")
+            except (OSError, UnicodeDecodeError, ValueError):
+                continue
+            name = "/" + relative.strip("/")
+            if not re.fullmatch(r"/[A-Za-z0-9][A-Za-z0-9_\-/]*", name):
+                continue
+            description, prompt = parse_custom_command(body)
+            commands.setdefault(
+                name,
+                {
+                    "description": description or first_non_empty_line(prompt) or "run saved prompt",
+                    "path": str(path),
+                    "prompt": prompt,
+                    "scope": scope,
+                },
+            )
+    return commands
+
+
+def custom_command_directories(root: Path) -> list[tuple[Path, str]]:
+    return [
+        (root / ".akernel" / "commands", "project"),
+        (Path.home() / ".akernel" / "commands", "user"),
+    ]
+
+
+def parse_custom_command(text: str) -> tuple[str, str]:
+    description = ""
+    body = text.strip()
+    if body.startswith("---"):
+        parts = body.split("---", 2)
+        if len(parts) == 3:
+            meta = parts[1]
+            body = parts[2].strip()
+            for line in meta.splitlines():
+                key, _, value = line.partition(":")
+                if key.strip().casefold() == "description":
+                    description = value.strip().strip('"').strip("'")
+                    break
+    return description, body
+
+
+def first_non_empty_line(text: str) -> str:
+    for line in text.splitlines():
+        value = line.strip().lstrip("#").strip()
+        if value:
+            return value[:80]
+    return ""
+
+
+def print_custom_commands_panel(root: Path) -> None:
+    commands = load_custom_chat_commands(root)
+    if not commands:
+        chat_notice(
+            "Slash Commands",
+            "No custom commands yet. Add Markdown prompts under .akernel/commands or ~/.akernel/commands.",
+        )
+        return
+    chat_panel(
+        "Slash Commands",
+        [
+            (name, f"{spec['description']} ({spec['scope']})")
+            for name, spec in sorted(commands.items())
+        ],
+    )
+
+
+def expand_custom_chat_command(root: Path, request: str) -> str | None:
+    command, _, arguments = request.strip().partition(" ")
+    if not command.startswith("/"):
+        return None
+    spec = load_custom_chat_commands(root).get(command)
+    if not spec:
+        return None
+    prompt = spec["prompt"]
+    if not prompt:
+        return arguments.strip() or f"Run custom command {command}."
+    arguments = arguments.strip()
+    return (
+        prompt.replace("{{args}}", arguments)
+        .replace("{{arguments}}", arguments)
+        .replace("$ARGUMENTS", arguments)
+        .strip()
+    )
+
+
 IGNORED_FILE_FINDER_DIRS = {
     ".git",
     ".hg",
@@ -1908,6 +2054,26 @@ IGNORED_FILE_FINDER_DIRS = {
 }
 
 
+def attach_inline_file_references(
+    workspace: Workspace,
+    tasks: TaskStore,
+    task_id: str,
+    request: str,
+    pending_context: list[str],
+) -> int:
+    attached = 0
+    seen: set[str] = set()
+    for match in INLINE_FILE_REF_RE.finditer(request):
+        path = match.group(1).strip("`'\".,;:!?)]}")
+        if not path or path.isdigit() or path in seen:
+            continue
+        seen.add(path)
+        if (workspace.root / path).is_file():
+            attach_chat_file(workspace, tasks, task_id, path, pending_context)
+            attached += 1
+    return attached
+
+
 def attach_chat_file_command(
     workspace: Workspace,
     tasks: TaskStore,

{akernel_runtime-0.1.8 → akernel_runtime-0.1.9}/tests/test_runtime.py

@@ -10,7 +10,14 @@ from unittest.mock import patch
 from context_kernel.agent_reports import build_agent_cost_report, render_agent_cost_report
 from context_kernel.benchmarks import BenchmarkRunner, render_benchmark_evidence_markdown, render_benchmark_markdown
 from context_kernel.budget import allocate_budget
-from context_kernel.cli import build_chat_tui_screen, chat_completion_items, load_batch_patch_specs, main, print_agent_report
+from context_kernel.cli import (
+    build_chat_tui_screen,
+    chat_completion_items,
+    expand_custom_chat_command,
+    load_batch_patch_specs,
+    main,
+    print_agent_report,
+)
 from context_kernel.context import ContextBuilder
 from context_kernel.evals import EvalRunner
 from context_kernel.global_memory import pull_global_memories, push_global_memories
@@ -1353,11 +1360,12 @@ class RuntimeTests(unittest.TestCase):
             reports = list(workspace.agent_runs_dir.glob("*.json"))
 
             self.assertEqual(len(reports), 1)
-            self.assertIn("AKERNEL // READY", output)
+            self.assertIn("AKERNEL", output)
+            self.assertIn("shortcuts /help /status /model /commands", output)
             self.assertIn("Assistant", output)
             self.assertIn("Mock agent response", output)
             self.assertIn("bye", output)
-            self.assertEqual(output.count("AKERNEL // READY"), 1)
+            self.assertEqual(output.count("shortcuts /help /status /model /commands"), 1)
 
     def test_tui_screen_can_render_older_history_window(self) -> None:
         with tempfile.TemporaryDirectory() as tmp:
@@ -1427,9 +1435,48 @@ class RuntimeTests(unittest.TestCase):
 
             command_items = chat_completion_items(root, "/mo")
             file_items = chat_completion_items(root, "@run")
+            inline_file_items = chat_completion_items(root, "please inspect @REA")
 
             self.assertIn(("/model", "show primary and auxiliary model roles"), command_items)
             self.assertIn(("@src/runtime.py", "attach file"), file_items)
+            self.assertIn(("@README.md", "attach file"), inline_file_items)
+
+    def test_chat_inline_file_reference_attaches_context(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            (root / "notes.md").write_text("inline context works", encoding="utf-8")
+            workspace = Workspace(root)
+            workspace.init()
+
+            with patch("builtins.input", side_effect=["Summarize @notes.md", "/exit"]):
+                with patch("sys.stdout", new=io.StringIO()):
+                    main(["--workspace", str(workspace.root), "chat", "--provider", "mock", "--max-steps", "1"])
+
+            reports = list(workspace.agent_runs_dir.glob("*.json"))
+            tool_traces = list(workspace.tool_traces_dir.glob("*.json"))
+            saved = Workspace.read_json(reports[0])
+
+            self.assertEqual(len(reports), 1)
+            self.assertGreaterEqual(len(tool_traces), 1)
+            self.assertIn("inline context works", saved["request"])
+
+    def test_custom_slash_command_expands_saved_prompt(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            command_dir = root / ".akernel" / "commands"
+            command_dir.mkdir(parents=True)
+            (command_dir / "review.md").write_text(
+                "---\ndescription: Review the current change\n---\nReview this change: {{args}}",
+                encoding="utf-8",
+            )
+            workspace = Workspace(root)
+            workspace.init()
+
+            items = chat_completion_items(root, "/rev")
+            expanded = expand_custom_chat_command(root, "/review cli polish")
+
+            self.assertIn(("/review", "project command: Review the current change"), items)
+            self.assertEqual(expanded, "Review this change: cli polish")
 
     def test_tui_screen_surfaces_task_plan_and_command_strip(self) -> None:
         with tempfile.TemporaryDirectory() as tmp: