1bcoder 0.1.10__tar.gz → 0.1.11__tar.gz

{1bcoder-0.1.10 → 1bcoder-0.1.11}/1bcoder.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: 1bcoder
-Version: 0.1.10
+Version: 0.1.11
 Summary: AI coding assistant agent for 1B–7B local models (Ollama, LMStudio, llama.cpp). Terminal REPL with file editing, project map, agents, scripts, and parallel multi-model queries.
 Project-URL: Homepage, https://github.com/szholobetsky/1bcoder
 Project-URL: Repository, https://github.com/szholobetsky/1bcoder
@@ -13,6 +13,8 @@ Requires-Dist: tqdm>=4.64
 Requires-Dist: rich>=13.0
 Dynamic: license-file
 
+![1bcoder](images/1bcoder.png)
+
 # 1bcoder
 
 AI coding assistant for small local models (0.5B–4B) — runs via [Ollama](https://ollama.com), [LMStudio](https://lmstudio.ai), or any OpenAI-compatible backend.
@@ -128,7 +130,7 @@ Tasks that require the model to decide *what to look at* — refactoring across
 - **Command autocorrection** — typos in command names, file paths, and keywords are detected and fixed automatically before execution, for both human input and agent actions
 - **`/tree [path]`** — display directory tree of the whole project or any subtree; ask to inject into context (or pass `ctx` to skip the prompt)
 - **`/find <pattern>`** — search filenames and file content with regex; supports `-f`/`-c`/`-i`/`--ext` flags; highlights matches, asks to inject results into context; sets `{{find_files}}` after every search; **`/find <terms> -r`** ranked BM25 mode returns top-10 files by relevance; hidden directories (`.git`, `.venv`, etc.) excluded automatically
-- AI proposes a **one-line fix** (`/fix`) or a **SEARCH/REPLACE patch** (`/patch`) — always shows a diff before applying
+- AI proposes a **one-line fix** (`/fix`), a **SEARCH/REPLACE patch** (`/patch`), or a **FIM-based fix** (`/fim`) — always shows a unified diff before applying
 - **Apply AI code blocks directly** with `/edit <file> code` (new/full file) or `/patch <file> code` (SEARCH/REPLACE from reply, no line numbers needed) — preferred for agent mode
 - **`<think>` tag support** — reasoning blocks shown in terminal by default; `/think hide` suppresses terminal display; `/think include` keeps reasoning in context for chained turns
 - Run shell commands and inject their output with `/run`
@@ -453,10 +455,12 @@ Then configure `/translate` to use it:
 | Command | Description |
 |---|---|
 | `/fix <file> [start-end] [hint]` | AI proposes one-line fix, shows diff, asks to apply |
+| `/fim <file> <line or start-end> [-w N] [hint]` | FIM-based fix — model rewrites the marked section, shows unified diff |
 | `/patch <file> [start-end] [hint]` | AI proposes SEARCH/REPLACE block, shows unified diff |
 | `/patch <file> code` | Apply SEARCH/REPLACE block from last AI reply (no new LLM call) |
 
-`/fix` is designed for 1B models — output is strictly constrained to `LINE N: content`.
+`/fix` asks the model to output `LINE N: content` — simple but small models often lose indentation.
+`/fim` (Fill-In-the-Middle) marks the target line(s) with `<<<...>>>` inside the full file, asks the model to return the corrected file, then diffs the result — no output format to learn, grammar constraints come from the surrounding code. Use `-w N` to limit context to N lines around the target when the file is larger than the context window.
 `/patch` works better with larger models (7B+) and can replace multiple consecutive lines.
 `/patch <file> code` is the preferred agent mode edit — the agent writes the SEARCH/REPLACE block in its reply, then calls `/patch <file> code` to apply it without needing line numbers.
 
@@ -465,6 +469,9 @@ When `/patch` fails to find the SEARCH text it shows a diagnostic diff — the S
 ```
 /fix main.py
 /fix main.py 2-2 wrong operator
+/fim main.py 3 replace != with ==
+/fim main.py 3-5 fix the logic
+/fim huge_file.py 14567 -w 20 fix indentation
 /patch main.py 10-40 fix the loop logic
 /patch main.py code
 ```
@@ -1739,7 +1746,7 @@ On Windows: hold `Shift` and drag with the left mouse button to select and copy
 ## Tips for 1B models
 
 - **Start small.** Use `/read file.py 10-25` instead of loading the whole file. Short context = better focus.
-- **Use `/fix` not `/patch`.** The `LINE N: content` format is much more reliable at 1B scale than free-form generation.
+- **Try `/fim` for tricky fixes.** FIM-based editing activates the model's grammar knowledge through surrounding code context — more reliable than `/fix` for indentation and multi-line logic. Use `-w 20` for large files.
 - **Build a map first.** Run `/map index .` at the start of a session, then use `/map find` to load only the relevant parts into context.
 - **Use scripts.** Scripts make multi-step work reproducible — the model only needs to handle one step at a time.
 - **Capture workflows.** After solving a task manually, run `/script create ctx` to save the exact steps as a reusable script.
@@ -1796,6 +1803,26 @@ For human input, the corrected command is shown with `[fix?]` and you are asked
 
 ---
 
+## Part of the SIMARGL toolkit
+
+1bcoder is one of four tools that together form an **intellectual development support system**:
+
+| Tool | Role |
+|---|---|
+| **[simargl](https://github.com/szholobetsky/simargl)** | Task-to-code retrieval — given a task description, finds which files and modules are likely affected, using semantic similarity over git history |
+| **[svitovyd](https://github.com/szholobetsky/svitovyd)** | Project map — scans any codebase and produces a structural map of definitions and cross-file dependencies; exposes it as an MCP server |
+| **[1bcoder](https://github.com/szholobetsky/1bcoder)** | AI coding assistant for small local models — surgical context management, agents, parallel inference, proc scripts |
+| **[yasna](https://github.com/szholobetsky/yasna)** | Session memory — indexes conversations from all AI agents so you can find what was discussed, when, and where |
+
+- **simargl** answers: *what code is related to this task?*
+- **svitovyd** answers: *how is the code structured and what depends on what?*
+- **1bcoder** answers: *how do I work with local models efficiently?*
+- **yasna** answers: *where did I already discuss this?*
+
+Together they cover the full development loop: understand the codebase, find relevant history, work with AI locally, remember what was decided.
+
+---
+
 **(c) 2026 Stanislav Zholobetskyi**
 Institute for Information Recording, National Academy of Sciences of Ukraine, Kyiv
 *PhD research: «Intelligent Technology for Software Development and Maintenance Support»*

{1bcoder-0.1.10 → 1bcoder-0.1.11}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: 1bcoder
-Version: 0.1.10
+Version: 0.1.11
 Summary: AI coding assistant agent for 1B–7B local models (Ollama, LMStudio, llama.cpp). Terminal REPL with file editing, project map, agents, scripts, and parallel multi-model queries.
 Project-URL: Homepage, https://github.com/szholobetsky/1bcoder
 Project-URL: Repository, https://github.com/szholobetsky/1bcoder
@@ -13,6 +13,8 @@ Requires-Dist: tqdm>=4.64
 Requires-Dist: rich>=13.0
 Dynamic: license-file
 
+![1bcoder](images/1bcoder.png)
+
 # 1bcoder
 
 AI coding assistant for small local models (0.5B–4B) — runs via [Ollama](https://ollama.com), [LMStudio](https://lmstudio.ai), or any OpenAI-compatible backend.
@@ -128,7 +130,7 @@ Tasks that require the model to decide *what to look at* — refactoring across
 - **Command autocorrection** — typos in command names, file paths, and keywords are detected and fixed automatically before execution, for both human input and agent actions
 - **`/tree [path]`** — display directory tree of the whole project or any subtree; ask to inject into context (or pass `ctx` to skip the prompt)
 - **`/find <pattern>`** — search filenames and file content with regex; supports `-f`/`-c`/`-i`/`--ext` flags; highlights matches, asks to inject results into context; sets `{{find_files}}` after every search; **`/find <terms> -r`** ranked BM25 mode returns top-10 files by relevance; hidden directories (`.git`, `.venv`, etc.) excluded automatically
-- AI proposes a **one-line fix** (`/fix`) or a **SEARCH/REPLACE patch** (`/patch`) — always shows a diff before applying
+- AI proposes a **one-line fix** (`/fix`), a **SEARCH/REPLACE patch** (`/patch`), or a **FIM-based fix** (`/fim`) — always shows a unified diff before applying
 - **Apply AI code blocks directly** with `/edit <file> code` (new/full file) or `/patch <file> code` (SEARCH/REPLACE from reply, no line numbers needed) — preferred for agent mode
 - **`<think>` tag support** — reasoning blocks shown in terminal by default; `/think hide` suppresses terminal display; `/think include` keeps reasoning in context for chained turns
 - Run shell commands and inject their output with `/run`
@@ -453,10 +455,12 @@ Then configure `/translate` to use it:
 | Command | Description |
 |---|---|
 | `/fix <file> [start-end] [hint]` | AI proposes one-line fix, shows diff, asks to apply |
+| `/fim <file> <line or start-end> [-w N] [hint]` | FIM-based fix — model rewrites the marked section, shows unified diff |
 | `/patch <file> [start-end] [hint]` | AI proposes SEARCH/REPLACE block, shows unified diff |
 | `/patch <file> code` | Apply SEARCH/REPLACE block from last AI reply (no new LLM call) |
 
-`/fix` is designed for 1B models — output is strictly constrained to `LINE N: content`.
+`/fix` asks the model to output `LINE N: content` — simple but small models often lose indentation.
+`/fim` (Fill-In-the-Middle) marks the target line(s) with `<<<...>>>` inside the full file, asks the model to return the corrected file, then diffs the result — no output format to learn, grammar constraints come from the surrounding code. Use `-w N` to limit context to N lines around the target when the file is larger than the context window.
 `/patch` works better with larger models (7B+) and can replace multiple consecutive lines.
 `/patch <file> code` is the preferred agent mode edit — the agent writes the SEARCH/REPLACE block in its reply, then calls `/patch <file> code` to apply it without needing line numbers.
 
@@ -465,6 +469,9 @@ When `/patch` fails to find the SEARCH text it shows a diagnostic diff — the S
 ```
 /fix main.py
 /fix main.py 2-2 wrong operator
+/fim main.py 3 replace != with ==
+/fim main.py 3-5 fix the logic
+/fim huge_file.py 14567 -w 20 fix indentation
 /patch main.py 10-40 fix the loop logic
 /patch main.py code
 ```
@@ -1739,7 +1746,7 @@ On Windows: hold `Shift` and drag with the left mouse button to select and copy
 ## Tips for 1B models
 
 - **Start small.** Use `/read file.py 10-25` instead of loading the whole file. Short context = better focus.
-- **Use `/fix` not `/patch`.** The `LINE N: content` format is much more reliable at 1B scale than free-form generation.
+- **Try `/fim` for tricky fixes.** FIM-based editing activates the model's grammar knowledge through surrounding code context — more reliable than `/fix` for indentation and multi-line logic. Use `-w 20` for large files.
 - **Build a map first.** Run `/map index .` at the start of a session, then use `/map find` to load only the relevant parts into context.
 - **Use scripts.** Scripts make multi-step work reproducible — the model only needs to handle one step at a time.
 - **Capture workflows.** After solving a task manually, run `/script create ctx` to save the exact steps as a reusable script.
@@ -1796,6 +1803,26 @@ For human input, the corrected command is shown with `[fix?]` and you are asked
 
 ---
 
+## Part of the SIMARGL toolkit
+
+1bcoder is one of four tools that together form an **intellectual development support system**:
+
+| Tool | Role |
+|---|---|
+| **[simargl](https://github.com/szholobetsky/simargl)** | Task-to-code retrieval — given a task description, finds which files and modules are likely affected, using semantic similarity over git history |
+| **[svitovyd](https://github.com/szholobetsky/svitovyd)** | Project map — scans any codebase and produces a structural map of definitions and cross-file dependencies; exposes it as an MCP server |
+| **[1bcoder](https://github.com/szholobetsky/1bcoder)** | AI coding assistant for small local models — surgical context management, agents, parallel inference, proc scripts |
+| **[yasna](https://github.com/szholobetsky/yasna)** | Session memory — indexes conversations from all AI agents so you can find what was discussed, when, and where |
+
+- **simargl** answers: *what code is related to this task?*
+- **svitovyd** answers: *how is the code structured and what depends on what?*
+- **1bcoder** answers: *how do I work with local models efficiently?*
+- **yasna** answers: *where did I already discuss this?*
+
+Together they cover the full development loop: understand the codebase, find relevant history, work with AI locally, remember what was decided.
+
+---
+
 **(c) 2026 Stanislav Zholobetskyi**
 Institute for Information Recording, National Academy of Sciences of Ukraine, Kyiv
 *PhD research: «Intelligent Technology for Software Development and Maintenance Support»*

{1bcoder-0.1.10 → 1bcoder-0.1.11}/README.md

@@ -1,3 +1,5 @@
+![1bcoder](images/1bcoder.png)
+
 # 1bcoder
 
 AI coding assistant for small local models (0.5B–4B) — runs via [Ollama](https://ollama.com), [LMStudio](https://lmstudio.ai), or any OpenAI-compatible backend.
@@ -113,7 +115,7 @@ Tasks that require the model to decide *what to look at* — refactoring across
 - **Command autocorrection** — typos in command names, file paths, and keywords are detected and fixed automatically before execution, for both human input and agent actions
 - **`/tree [path]`** — display directory tree of the whole project or any subtree; ask to inject into context (or pass `ctx` to skip the prompt)
 - **`/find <pattern>`** — search filenames and file content with regex; supports `-f`/`-c`/`-i`/`--ext` flags; highlights matches, asks to inject results into context; sets `{{find_files}}` after every search; **`/find <terms> -r`** ranked BM25 mode returns top-10 files by relevance; hidden directories (`.git`, `.venv`, etc.) excluded automatically
-- AI proposes a **one-line fix** (`/fix`) or a **SEARCH/REPLACE patch** (`/patch`) — always shows a diff before applying
+- AI proposes a **one-line fix** (`/fix`), a **SEARCH/REPLACE patch** (`/patch`), or a **FIM-based fix** (`/fim`) — always shows a unified diff before applying
 - **Apply AI code blocks directly** with `/edit <file> code` (new/full file) or `/patch <file> code` (SEARCH/REPLACE from reply, no line numbers needed) — preferred for agent mode
 - **`<think>` tag support** — reasoning blocks shown in terminal by default; `/think hide` suppresses terminal display; `/think include` keeps reasoning in context for chained turns
 - Run shell commands and inject their output with `/run`
@@ -438,10 +440,12 @@ Then configure `/translate` to use it:
 | Command | Description |
 |---|---|
 | `/fix <file> [start-end] [hint]` | AI proposes one-line fix, shows diff, asks to apply |
+| `/fim <file> <line or start-end> [-w N] [hint]` | FIM-based fix — model rewrites the marked section, shows unified diff |
 | `/patch <file> [start-end] [hint]` | AI proposes SEARCH/REPLACE block, shows unified diff |
 | `/patch <file> code` | Apply SEARCH/REPLACE block from last AI reply (no new LLM call) |
 
-`/fix` is designed for 1B models — output is strictly constrained to `LINE N: content`.
+`/fix` asks the model to output `LINE N: content` — simple but small models often lose indentation.
+`/fim` (Fill-In-the-Middle) marks the target line(s) with `<<<...>>>` inside the full file, asks the model to return the corrected file, then diffs the result — no output format to learn, grammar constraints come from the surrounding code. Use `-w N` to limit context to N lines around the target when the file is larger than the context window.
 `/patch` works better with larger models (7B+) and can replace multiple consecutive lines.
 `/patch <file> code` is the preferred agent mode edit — the agent writes the SEARCH/REPLACE block in its reply, then calls `/patch <file> code` to apply it without needing line numbers.
 
@@ -450,6 +454,9 @@ When `/patch` fails to find the SEARCH text it shows a diagnostic diff — the S
 ```
 /fix main.py
 /fix main.py 2-2 wrong operator
+/fim main.py 3 replace != with ==
+/fim main.py 3-5 fix the logic
+/fim huge_file.py 14567 -w 20 fix indentation
 /patch main.py 10-40 fix the loop logic
 /patch main.py code
 ```
@@ -1724,7 +1731,7 @@ On Windows: hold `Shift` and drag with the left mouse button to select and copy
 ## Tips for 1B models
 
 - **Start small.** Use `/read file.py 10-25` instead of loading the whole file. Short context = better focus.
-- **Use `/fix` not `/patch`.** The `LINE N: content` format is much more reliable at 1B scale than free-form generation.
+- **Try `/fim` for tricky fixes.** FIM-based editing activates the model's grammar knowledge through surrounding code context — more reliable than `/fix` for indentation and multi-line logic. Use `-w 20` for large files.
 - **Build a map first.** Run `/map index .` at the start of a session, then use `/map find` to load only the relevant parts into context.
 - **Use scripts.** Scripts make multi-step work reproducible — the model only needs to handle one step at a time.
 - **Capture workflows.** After solving a task manually, run `/script create ctx` to save the exact steps as a reusable script.
@@ -1781,6 +1788,26 @@ For human input, the corrected command is shown with `[fix?]` and you are asked
 
 ---
 
+## Part of the SIMARGL toolkit
+
+1bcoder is one of four tools that together form an **intellectual development support system**:
+
+| Tool | Role |
+|---|---|
+| **[simargl](https://github.com/szholobetsky/simargl)** | Task-to-code retrieval — given a task description, finds which files and modules are likely affected, using semantic similarity over git history |
+| **[svitovyd](https://github.com/szholobetsky/svitovyd)** | Project map — scans any codebase and produces a structural map of definitions and cross-file dependencies; exposes it as an MCP server |
+| **[1bcoder](https://github.com/szholobetsky/1bcoder)** | AI coding assistant for small local models — surgical context management, agents, parallel inference, proc scripts |
+| **[yasna](https://github.com/szholobetsky/yasna)** | Session memory — indexes conversations from all AI agents so you can find what was discussed, when, and where |
+
+- **simargl** answers: *what code is related to this task?*
+- **svitovyd** answers: *how is the code structured and what depends on what?*
+- **1bcoder** answers: *how do I work with local models efficiently?*
+- **yasna** answers: *where did I already discuss this?*
+
+Together they cover the full development loop: understand the codebase, find relevant history, work with AI locally, remember what was decided.
+
+---
+
 **(c) 2026 Stanislav Zholobetskyi**
 Institute for Information Recording, National Academy of Sciences of Ukraine, Kyiv
 *PhD research: «Intelligent Technology for Software Development and Maintenance Support»*

{1bcoder-0.1.10 → 1bcoder-0.1.11}/_bcoder_data/agents/advance.txt

@@ -13,6 +13,7 @@ system =
 
     How to write files:
     - To MODIFY an existing file: write a SEARCH/REPLACE block, then ACTION: /patch <file> code
+      Or use FIM for targeted fixes: ACTION: /fim <file> <line or start-end> <hint>
     - To INSERT new code at a line: write the code block, then ACTION: /insert <file> <line> code
     - To CREATE or fully REPLACE a file: write the full code block, then ACTION: /save <file> code
 
@@ -24,7 +25,7 @@ system =
     >>>>>>> REPLACE
 
     Rules:
-    - /read a file before editing it
+    - /readln a file before editing it
     - /bkup save <file> before modifying important files
     - /run to test after applying a fix
 
@@ -40,7 +41,7 @@ tools =
     bkup
     diff
     patch
-    fix
+    fim
     tree
     find
     map

{1bcoder-0.1.10 → 1bcoder-0.1.11}/_bcoder_data/doc/PARAM.md

@@ -17,6 +17,8 @@ These are handled by 1bcoder directly and are not forwarded to the model API.
 | `think_exclude` | `true` | Strip `<think>…</think>` blocks from context (blocks are shown in terminal only). Set `false` to keep reasoning in context for multi-turn chains. |
 | `ask_limit` | 8000 | Max characters of `/ask` tool output kept in context. |
 | `ask_show` | 500 | Characters shown in terminal when output is truncated. |
+| `run_timeout` | 30 | Timeout in seconds for `/run` shell commands. Set to `0` to disable (no timeout). Useful for long-running CLI tools like `simargl search`. |
+| `log` | `false` | Print full request details before each LLM call: URL, model, message count, options. Also prints HTTP response headers on error. Useful for debugging 500 errors or unexpected model behaviour. |
 
 ---
 

{1bcoder-0.1.10 → 1bcoder-0.1.11}/chat.py

@@ -225,7 +225,7 @@ PROFILES_FILE        = os.path.join(BCODER_DIR, "profiles.txt")
 GLOBAL_PROFILES_FILE = os.path.join(HOME_BCODER_DIR, "profiles.txt")
 
 DEFAULT_AGENT_TOOLS = [
-    "read", "tree", "find", "insert", "save", "patch",
+    "read", "readln", "tree", "find", "insert", "save", "fim",
 ]
 
 DEFAULT_AGENT_TOOLS_ADVANCED = [
@@ -247,33 +247,19 @@ AGENT_SYSTEM_BASIC = """\
 You are a coding assistant. Complete the task using the available tools.
 
 To call a tool, write ACTION: on its own line. Wait for [tool result].
-Run actions in a loop until the task is complete.
-When all actions are done, write a short summary with no ACTION and end with: Task is complete.
+Run one action at a time until the task is complete.
+When done, write a short summary with no ACTION and end with: Task is complete.
 
-Available actions:
-  ACTION: /read <file>            ← read whole file
-  ACTION: /read <file> 35-55     ← read lines 35–55
+To fix or edit a file:
+  ACTION: /fim <file> <description of what to fix>
+  Example: ACTION: /fim calc.py fix the division by zero bug
+  Example: ACTION: /fim main.py fix the indentation on line 5
 
-To modify a file:
-1. Write the code block (```...```)
-2. Then call:
-   ACTION: /insert <file> <line> code        ← insert code block before line N
-   ACTION: /insert <file> <line> <text>      ← insert literal text before line N
-   ACTION: /patch <file> code                ← apply SEARCH/REPLACE block
-   ACTION: /save <file> code                 ← save or overwrite whole file
+To read a file:
+  ACTION: /read <file>
 
-The keyword "code" in the action means: use the last code block (``` ```) written above as the content.
-
-SEARCH/REPLACE format for /patch:
-<<<<<<< SEARCH
-exact lines to replace
-=======
-new lines
->>>>>>> REPLACE
-
-Rules:
-- If you don't know the project structure, start with ACTION: /tree
-- Always /read a file before inserting or patching it.
+To create or fully replace a file: write the full code block, then:
+  ACTION: /save <file> code
 
 Available tools:
 {tool_list}
@@ -287,6 +273,7 @@ When the task is complete, write a short summary and end with: Task is complete.
 
 How to write files (the keyword "code" means: use the last ``` ``` block written above):
 - To MODIFY an existing file: write a SEARCH/REPLACE block, then ACTION: /patch <file> code
+  Or use FIM for targeted fixes: ACTION: /fim <file> <line or start-end> <hint>
 - To INSERT new code at a line: write the code block, then ACTION: /insert <file> <line> code
 - To CREATE or fully REPLACE a file: write the full code block, then ACTION: /save <file> code
 
@@ -298,7 +285,7 @@ new lines
 >>>>>>> REPLACE
 
 Rules:
-- /read a file before editing it
+- /readln a file before editing it
 - /bkup save <file> before modifying important files
 - /run to test after applying a fix
 
@@ -339,7 +326,7 @@ FIX_SYSTEM = (
     "You are a code repair tool. "
     "Respond with ONLY the single most important fix in this exact format:\n"
     "LINE <number>: <corrected line content>\n"
-    "One fix only. No explanation. No other text. Preserve indentation."
+    "One fix only. No explanation. No other text."
 )
 
 PATCH_SYSTEM = (
@@ -450,6 +437,16 @@ Commands
           /fix main.py 2-2
           /fix main.py 2-2 wrong operator
 
+/fim <file> [line or start-end] [-w N] [hint]
+    FIM-based fix. Marks the line(s), passes file to model, diffs the result.
+    Without a line number: passes the whole file with hint only (agent fallback mode).
+    -w N  pass only N lines of context around the marked section (for large files).
+    e.g.  /fim main.py 3
+          /fim main.py 3 replace != with ==
+          /fim main.py 3-5 fix the logic
+          /fim huge.py 14567 -w 20 fix indentation
+          /fim main.py fix the division by zero bug
+
 /patch <file> [start-end] [hint]
     AI proposes a multi-line SEARCH/REPLACE edit. Shows unified diff before applying.
     Better for 7B+ models. Use /fix for 1B models.
@@ -874,6 +871,7 @@ Output capture operators (work with any command — LLM reply, tool, proc):
       \\!term exclude entire block if any child contains term
       -term   show ONLY child lines containing term
       -!term  hide child lines containing term
+      +term   like -term but also hides files with no matching children
     e.g.  /map find register
           /map find \\register !mock
           /map find auth \\UserService -!deprecated -y
@@ -1271,6 +1269,54 @@ def ai_fix(base_url, model, content, label, hint="", on_chunk=None, provider="ol
     return None, raw
 
 
+def ai_fill(base_url, model, file_lines, start, end, hint="", on_chunk=None, provider="ollama"):
+    """FIM-based fix: marks line range in the whole file, asks model for corrected file.
+    If start is None: no marker — whole file passed with hint only (agent fallback mode)."""
+    if start is None:
+        file_content = "".join(file_lines)
+        user_msg = file_content + "\nFix the issue. Output the complete corrected file."
+    else:
+        marked = list(file_lines)
+        if start == end:
+            marked[start - 1] = f"<<<{file_lines[start - 1].rstrip()}>>>\n"
+        else:
+            marked[start - 1] = "<<<\n" + file_lines[start - 1]
+            marked[end - 1]   = file_lines[end - 1].rstrip() + "\n>>>\n"
+        file_content = "".join(marked)
+        user_msg = file_content + "\nFix the <<<...>>> section. Output the complete corrected file."
+    if hint:
+        user_msg = f"{hint}\n\n{user_msg}"
+    msgs = [{"role": "user", "content": user_msg}]
+    chunks = []
+    if provider == "openai":
+        with requests.post(
+            f"{base_url}/v1/chat/completions",
+            json={"model": model, "messages": msgs, "stream": True},
+            stream=True, timeout=120,
+        ) as resp:
+            resp.raise_for_status()
+            _parse_openai_stream(resp, on_chunk, chunks)
+    else:
+        with requests.post(
+            f"{base_url}/api/chat",
+            json={"model": model, "messages": msgs, "stream": True},
+            stream=True, timeout=120,
+        ) as resp:
+            resp.raise_for_status()
+            for line in resp.iter_lines():
+                if not line:
+                    continue
+                data = json.loads(line)
+                chunk = data.get("message", {}).get("content", "")
+                if chunk:
+                    if on_chunk:
+                        on_chunk(chunk)
+                    chunks.append(chunk)
+                if data.get("done"):
+                    break
+    return "".join(chunks)
+
+
 def _parse_patch(text):
     """Extract (search_text, replace_text) from a SEARCH/REPLACE block, or (None, None)."""
     m = re.search(
@@ -1984,7 +2030,7 @@ def _map_patch_remove(map_path: str, rel_prefix: str) -> int:
 
 _KNOWN_CMDS = [
     "/read", "/readln", "/insert", "/edit", "/save", "/run", "/script", "/mcp",
-    "/parallel", "/patch", "/fix", "/bkup", "/diff", "/agent", "/tree",
+    "/parallel", "/patch", "/fix", "/fim", "/bkup", "/diff", "/agent", "/tree",
     "/find", "/map", "/ctx", "/think", "/format", "/param", "/model",
     "/host", "/help", "/init", "/clear", "/exit",
     "/prompt", "/proc", "/team", "/var", "/config", "/alias",
@@ -2326,6 +2372,7 @@ class CoderCLI:
         self._last_find_results: list = []  # numbered results from last /proj find
         self._history: list[str] = []
         self.cmd_history: list[str] = []   # all /commands typed this session
+        self.log_requests: bool = False    # /param log true — print request body + error detail
         # enable readline history if available
         try:
             import readline
@@ -2376,6 +2423,9 @@ class CoderCLI:
             if self.provider == "openai":
                 body = {"model": self.model, "messages": messages, "stream": True}
                 body.update(self.params)
+                if self.log_requests:
+                    print(f"  [log] POST {self.base_url}/v1/chat/completions")
+                    print(f"  [log] model={self.model} messages={len(messages)} params={self.params}")
                 with requests.post(
                     f"{self.base_url}/v1/chat/completions",
                     json=body, stream=True, timeout=self.timeout,
@@ -2436,6 +2486,9 @@ class CoderCLI:
             else:
                 opts = {"num_ctx": self.num_ctx}
                 opts.update(self.params)
+                if self.log_requests:
+                    print(f"  [log] POST {self.base_url}/api/chat")
+                    print(f"  [log] model={self.model} messages={len(messages)} options={opts}")
                 with requests.post(
                     f"{self.base_url}/api/chat",
                     json={"model": self.model, "messages": messages, "stream": True,
@@ -2494,6 +2547,13 @@ class CoderCLI:
             return None  # sentinel: interrupted (vs "" which means empty reply)
         except requests.exceptions.RequestException as e:
             print(f"\nerror: {e}")
+            if hasattr(e, "response") and e.response is not None:
+                body = e.response.text.strip()
+                if body:
+                    print(f"  detail: {body[:500]}")
+                if self.log_requests:
+                    print(f"  [log] status: {e.response.status_code}")
+                    print(f"  [log] headers: {dict(e.response.headers)}")
             return ""
         print()
         reply = "".join(chunks)
@@ -2706,6 +2766,8 @@ class CoderCLI:
             if self._run_hook("before_patch", _ha):
                 self._cmd_patch(user_input)
                 self._run_hook("after_patch", _ha)
+        elif user_input.startswith("/fim"):
+            self._cmd_fill(user_input)
         elif user_input.startswith("/fix"):
             _ha = self._extract_hook_args(user_input)
             if self._run_hook("before_fix", _ha):
@@ -2882,6 +2944,8 @@ advanced_tools =
             print(f"  num_ctx        = {self.num_ctx}")
             print(f"  think_exclude  = {not self.think_in_ctx}  (strip <think> blocks from context)")
             print(f"  think_show     = {self.think_show}  (show <think> blocks in terminal)")
+            print(f"  log            = {self.log_requests}  (print request body + error detail)")
+            print(f"  run_timeout    = {self.params.get('run_timeout', 30)}s  (0 = no timeout, for /run commands)")
             print(f"  ask_limit      = {self.params.get('ask_limit', ASK_RESULT_LIMIT_CHARS)}  (chars, /ask truncation limit)")
             print(f"  ask_show       = {self.params.get('ask_show',  ASK_RESULT_SHOW_CHARS)}  (chars shown when truncated)")
             model_params = {k: v for k, v in self.params.items() if k not in ("ask_limit", "ask_show")}
@@ -2927,6 +2991,10 @@ advanced_tools =
             self.think_show = bool(val)
             _ok(f"[param] think_show = {self.think_show}")
             return
+        if key == "log":
+            self.log_requests = bool(val)
+            _ok(f"[param] log = {self.log_requests}")
+            return
         self.params[key] = val
         _ok(f"[param] {key} = {val}")
 
@@ -4061,7 +4129,7 @@ advanced_tools =
             return
         try:
             current_text, _ = read_file(path, lineno, lineno)
-            current_text = current_text.split(":", 1)[1].rstrip("\n") if ":" in current_text else current_text.rstrip()
+            current_text = current_text.split(":", 1)[1][1:].rstrip("\n") if ":" in current_text else current_text.rstrip()
             print(f"  current [{lineno}]: {current_text}")
             print(f"  new     [{lineno}]: {new_content}")
         except (FileNotFoundError, OSError):
@@ -4075,6 +4143,95 @@ advanced_tools =
         else:
             print("[skipped]")
 
+    def _cmd_fill(self, user_input: str):
+        parts = user_input[4:].strip().split(None, 2)
+        path = parts[0] if parts else ""
+        start = end = None
+        hint = ""
+        if not path:
+            print("usage: /fim <file> <line or start-end> [-w N] [hint]")
+            return
+        if len(parts) >= 2:
+            m = re.match(r'^(\d+)(?:-(\d+))?$', parts[1])
+            if m:
+                start = int(m.group(1))
+                end   = int(m.group(2)) if m.group(2) else start
+                hint  = parts[2] if len(parts) >= 3 else ""
+            else:
+                hint = " ".join(parts[1:])
+        if start is None and not hint:
+            print("usage: /fim <file> [line or start-end] [-w N] [hint]")
+            return
+        # extract -w <window> from hint
+        window = None
+        wm = re.search(r'(?:^|\s)-w\s+(\d+)', hint)
+        if wm:
+            window = int(wm.group(1))
+            hint = (hint[:wm.start()] + hint[wm.end():]).strip()
+        try:
+            with open(path, "r", encoding="utf-8") as f:
+                file_lines = f.readlines()
+        except (FileNotFoundError, OSError) as e:
+            _err(e)
+            return
+        total = len(file_lines)
+        if start is not None and (start < 1 or end > total):
+            _err(f"line {start}-{end} out of range (file has {total} lines)")
+            return
+        # compute slice for windowed mode (only when line range is given)
+        if start is not None and window is not None:
+            w_start = max(0, start - 1 - window)          # 0-indexed inclusive
+            w_end   = min(total, end + window)             # 0-indexed exclusive
+            slice_lines = file_lines[w_start:w_end]
+            rel_start   = start - w_start                  # 1-indexed within slice
+            rel_end     = end   - w_start
+        else:
+            slice_lines = file_lines
+            rel_start, rel_end = start, end
+            w_start, w_end = 0, total
+        if hint:
+            print(f"hint: {hint}")
+        if window is not None:
+            print(f"[fim] window: lines {w_start+1}–{w_end} of {total}")
+        self._sep("AI")
+        def on_chunk(c):
+            sys.stdout.write(c)
+            sys.stdout.flush()
+        try:
+            raw = ai_fill(self.base_url, self.model, slice_lines, rel_start, rel_end, hint, on_chunk, self.provider)
+        except KeyboardInterrupt:
+            print("\n[interrupted]")
+            return
+        except requests.exceptions.RequestException as e:
+            print(f"\nerror: {e}")
+            if hasattr(e, "response") and e.response is not None:
+                body = e.response.text.strip()
+                if body:
+                    print(f"  detail: {body[:500]}")
+            return
+        print()
+        m = re.search(r'```(?:\w+)?\n(.*?)\n```', raw, re.DOTALL)
+        new_content = m.group(1) if m else raw.strip()
+        new_slice = [l if l.endswith('\n') else l + '\n' for l in new_content.splitlines()]
+        # reconstruct full file: unchanged prefix + new slice + unchanged suffix
+        full_new = file_lines[:w_start] + new_slice + file_lines[w_end:]
+        diff = list(difflib.unified_diff(file_lines, full_new, fromfile=path, tofile=path + " (fixed)", lineterm=""))
+        if not diff:
+            print("[fim] no changes detected — try a different hint or model")
+            return
+        for dline in diff:
+            print(_cdiff(dline))
+        if self._confirm("  apply? [Y/n]:"):
+            try:
+                with open(path, "w", encoding="utf-8") as f:
+                    f.writelines(full_new)
+                changed = sum(1 for d in diff if d.startswith('+') and not d.startswith('+++'))
+                print(f"[{changed} line(s) updated in {path}]")
+            except OSError as e:
+                _err(e)
+        else:
+            print("[skipped]")
+
     # ── hook support ──────────────────────────────────────────────────────────
 
     def _extract_hook_args(self, user_input: str) -> dict:
@@ -4303,7 +4460,7 @@ advanced_tools =
         if search_text is None:
             print("could not parse SEARCH/REPLACE block — try a more capable model")
             return
-        if search_text.strip() == replace_text.strip():
+        if search_text == replace_text:
             _warn("[patch] SEARCH and REPLACE are identical — model included the new code in both blocks (no-op)")
             return
         try:
@@ -4454,9 +4611,11 @@ advanced_tools =
 
     def _cmd_run(self, shell_cmd: str):
         print(f"$ {shell_cmd}")
+        run_timeout = self.params.get("run_timeout", 30)
         try:
             proc = subprocess.run(
-                shell_cmd, shell=True, capture_output=True, timeout=30,
+                shell_cmd, shell=True, capture_output=True,
+                timeout=run_timeout if run_timeout else None,
                 encoding="utf-8", errors="replace",
             )
             output = proc.stdout + proc.stderr
@@ -7646,6 +7805,7 @@ Config stored in ~/.1bcoder/translate.json
             print("  /map find \\!term              — exclude block if any child contains term")
             print("  /map find -term                — show ONLY child lines containing term")
             print("  /map find -!term               — hide child lines containing term")
+            print("  /map find +term                — like -term but also hides files with no matching children")
             print("  combine freely: auth \\register !mock -!deprecated -y -d 2")
             print("  /map trace <identifier> [-d N] [-y]   — follow call chain backwards from identifier")
             print("  /map diff                      — diff map.txt vs map.prev.txt (no re-index)")
@@ -8275,13 +8435,7 @@ Config stored in ~/.1bcoder/translate.json
         if template is None:
             return user_input
         if "{{args}}" in template:
-            # Extract --flags separately; pass remaining args verbatim to preserve
-            # quoted segments and keywords like file: / plan: intact.
-            flags = re.findall(r'--\S+', args)
-            text  = re.sub(r'\s*--\S+', '', args).strip()
-            expanded = template.replace("{{args}}", text)
-            if flags:
-                expanded = expanded + " " + " ".join(flags)
+            expanded = template.replace("{{args}}", args)
         else:
             expanded = (template + (" " + args if args else "")).strip()
         return self._expand_alias(expanded, depth + 1)

{1bcoder-0.1.10 → 1bcoder-0.1.11}/map_query.py

@@ -22,6 +22,7 @@ Filter token syntax for find:
     \\!term exclude entire block if any child contains term
     -term   show ONLY child lines containing term
     -!term  hide child lines containing term
+    +term   show ONLY child lines containing term AND hide file if none match
 
     (no tokens → print full map)
 
@@ -154,13 +155,14 @@ def find_map(map_path: str, query: str) -> tuple:
     with open(map_path, encoding='utf-8') as f:
         content = f.read()
 
-    tokens     = query.split()
-    pos_file   = []   # term   — filename must contain
-    neg_file   = []   # !term  — filename must NOT contain
-    pos_child  = []   # \term  — include block if any child line contains (all terms same line)
-    neg_block  = []   # \!term — exclude block if any child line contains
-    show_lines = []   # -term  — show ONLY child lines containing term (whitelist)
-    hide_lines = []   # -!term — hide child lines containing term
+    tokens          = query.split()
+    pos_file        = []   # term   — filename must contain
+    neg_file        = []   # !term  — filename must NOT contain
+    pos_child       = []   # \term  — include block if any child line contains (all terms same line)
+    neg_block       = []   # \!term — exclude block if any child line contains
+    show_lines      = []   # -term  — show ONLY child lines containing term (keep file even if empty)
+    hide_lines      = []   # -!term — hide child lines containing term
+    must_show_lines = []   # +term  — show ONLY child lines containing term AND hide file if none match
 
     for t in tokens:
         if t.startswith('\\!') and len(t) > 2:
@@ -171,13 +173,15 @@ def find_map(map_path: str, query: str) -> tuple:
             hide_lines.append(t[2:].lower())
         elif t.startswith('-') and len(t) > 1:
             show_lines.append(t[1:].lower())
+        elif t.startswith('+') and len(t) > 1:
+            must_show_lines.append(t[1:].lower())
         elif t.startswith('!') and len(t) > 1:
             neg_file.append(t[1:].lower())
         else:
             pos_file.append(t.lower())
 
     # no criteria → return full map
-    if not any([pos_file, neg_file, pos_child, neg_block, show_lines, hide_lines]):
+    if not any([pos_file, neg_file, pos_child, neg_block, show_lines, hide_lines, must_show_lines]):
         return [], content
 
     blocks = re.split(r'\n(?=\S)', content)
@@ -201,6 +205,11 @@ def find_map(map_path: str, query: str) -> tuple:
         if show_lines:
             child_lines = [l for l in child_lines
                            if any(t in l.lower() for t in show_lines)]
+        if must_show_lines:
+            child_lines = [l for l in child_lines
+                           if any(t in l.lower() for t in must_show_lines)]
+            if not child_lines:
+                return None
         if hide_lines:
             child_lines = [l for l in child_lines
                            if not any(t in l.lower() for t in hide_lines)]

{1bcoder-0.1.10 → 1bcoder-0.1.11}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "1bcoder"
-version = "0.1.10"
+version = "0.1.11"
 description = "AI coding assistant agent for 1B–7B local models (Ollama, LMStudio, llama.cpp). Terminal REPL with file editing, project map, agents, scripts, and parallel multi-model queries."
 requires-python = ">=3.10"
 readme = {file = "README.md", content-type = "text/markdown"}