1bcoder 0.1.16__tar.gz → 0.1.18__tar.gz

{1bcoder-0.1.16 → 1bcoder-0.1.18}/1bcoder.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: 1bcoder
-Version: 0.1.16
+Version: 0.1.18
 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
@@ -1828,7 +1828,7 @@ For human input, the corrected command is shown with `[fix?]` and you are asked
 
 ## Part of the SIMARGL toolkit
 
-1bcoder is one of five tools that together form an **intellectual development support system**:
+1bcoder is one of six tools that together form an **intellectual development support system**:
 
 | Tool | Role |
 |---|---|
@@ -1837,12 +1837,14 @@ For human input, the corrected command is shown with `[fix?]` and you are asked
 | **[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 |
 | **[radogast](https://github.com/szholobetsky/radogast)** | Context drift monitor — measures how far an AI agent's conversation has drifted from the original task |
+| **[vyrii](https://github.com/szholobetsky/vyrii)** | Local AI web UI — chat, translate, web research, RAG, and file management via Gradio; powered by Ollama or any OpenAI-compatible backend |
 
 - **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?*
 - **radogast** answers: *is the AI agent still on track toward the goal?*
+- **vyrii** answers: *how do I access all of this through a browser?*
 
 Together they cover the full development loop: understand the codebase, find relevant history, work with AI locally, remember what was decided, and verify the context stays on target.
 

{1bcoder-0.1.16 → 1bcoder-0.1.18}/1bcoder.egg-info/SOURCES.txt

@@ -17,12 +17,15 @@ _bcoder_data/profiles.txt
 _bcoder_data/prompts.txt
 _bcoder_data/agents/advance.txt
 _bcoder_data/agents/ask.txt
+_bcoder_data/agents/coder.txt
+_bcoder_data/agents/codex.txt
 _bcoder_data/agents/compact.txt
 _bcoder_data/agents/concepts.txt
 _bcoder_data/agents/fill.txt
 _bcoder_data/agents/planning.txt
 _bcoder_data/agents/scan.txt
 _bcoder_data/agents/sqlite.txt
+_bcoder_data/agents/terminator.txt
 _bcoder_data/agents/websearch.txt
 _bcoder_data/doc/FLOWS.md
 _bcoder_data/doc/MCP.md
@@ -39,15 +42,23 @@ _bcoder_data/flows/external_help.py
 _bcoder_data/flows/grounding.py
 _bcoder_data/flows/obfuscate.py
 _bcoder_data/flows/py_error_trace.py
+_bcoder_data/flows/rag.py
+_bcoder_data/flows/scan.py
 _bcoder_data/flows/simargl_files.py
 _bcoder_data/flows/visual_search.py
+_bcoder_data/flows/webanalys.py
 _bcoder_data/flows/webask.py
 _bcoder_data/flows/webcrawl.py
+_bcoder_data/flows/webindex.py
 _bcoder_data/flows/__pycache__/commit_message.cpython-311.pyc
+_bcoder_data/flows/__pycache__/deepagent_md.cpython-311.pyc
+_bcoder_data/flows/__pycache__/rag.cpython-311.pyc
 _bcoder_data/flows/__pycache__/webcrawl.cpython-311.pyc
 _bcoder_data/proc/action-required.py
 _bcoder_data/proc/add-save.py
 _bcoder_data/proc/assist.py
+_bcoder_data/proc/autobkup.py
+_bcoder_data/proc/autocheck.py
 _bcoder_data/proc/collect-files.py
 _bcoder_data/proc/ctx_cut.py
 _bcoder_data/proc/ctx_expand.py
@@ -55,6 +66,7 @@ _bcoder_data/proc/extract-code.py
 _bcoder_data/proc/extract-files.py
 _bcoder_data/proc/extract-list.py
 _bcoder_data/proc/grounding-check.py
+_bcoder_data/proc/ladder.py
 _bcoder_data/proc/md.py
 _bcoder_data/proc/mdx.py
 _bcoder_data/proc/pattern-gate.py

{1bcoder-0.1.16 → 1bcoder-0.1.18}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: 1bcoder
-Version: 0.1.16
+Version: 0.1.18
 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
@@ -1828,7 +1828,7 @@ For human input, the corrected command is shown with `[fix?]` and you are asked
 
 ## Part of the SIMARGL toolkit
 
-1bcoder is one of five tools that together form an **intellectual development support system**:
+1bcoder is one of six tools that together form an **intellectual development support system**:
 
 | Tool | Role |
 |---|---|
@@ -1837,12 +1837,14 @@ For human input, the corrected command is shown with `[fix?]` and you are asked
 | **[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 |
 | **[radogast](https://github.com/szholobetsky/radogast)** | Context drift monitor — measures how far an AI agent's conversation has drifted from the original task |
+| **[vyrii](https://github.com/szholobetsky/vyrii)** | Local AI web UI — chat, translate, web research, RAG, and file management via Gradio; powered by Ollama or any OpenAI-compatible backend |
 
 - **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?*
 - **radogast** answers: *is the AI agent still on track toward the goal?*
+- **vyrii** answers: *how do I access all of this through a browser?*
 
 Together they cover the full development loop: understand the codebase, find relevant history, work with AI locally, remember what was decided, and verify the context stays on target.
 

{1bcoder-0.1.16 → 1bcoder-0.1.18}/README.md

@@ -1813,7 +1813,7 @@ For human input, the corrected command is shown with `[fix?]` and you are asked
 
 ## Part of the SIMARGL toolkit
 
-1bcoder is one of five tools that together form an **intellectual development support system**:
+1bcoder is one of six tools that together form an **intellectual development support system**:
 
 | Tool | Role |
 |---|---|
@@ -1822,12 +1822,14 @@ For human input, the corrected command is shown with `[fix?]` and you are asked
 | **[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 |
 | **[radogast](https://github.com/szholobetsky/radogast)** | Context drift monitor — measures how far an AI agent's conversation has drifted from the original task |
+| **[vyrii](https://github.com/szholobetsky/vyrii)** | Local AI web UI — chat, translate, web research, RAG, and file management via Gradio; powered by Ollama or any OpenAI-compatible backend |
 
 - **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?*
 - **radogast** answers: *is the AI agent still on track toward the goal?*
+- **vyrii** answers: *how do I access all of this through a browser?*
 
 Together they cover the full development loop: understand the codebase, find relevant history, work with AI locally, remember what was decided, and verify the context stays on target.
 

{1bcoder-0.1.16 → 1bcoder-0.1.18}/_bcoder_data/agents/advance.txt

@@ -1,5 +1,14 @@
-# Autonomous coding agent with full toolset — for larger models (7B+)
-# Usage: /advance <task>  or  /agent advance <task>
+# Autonomous coding agent with full toolset — for larger models (7B+).
+# auto_exec=false means every ACTION is shown for manual confirmation before running.
+# Has bkup tool: model can save backups itself before destructive edits.
+# Supports all edit strategies: save (full replace), patch (search/replace), fim (targeted fix), insert (line).
+# Use /readln before editing to get line numbers for patch/fim/insert.
+# Pair with gates: /proc gate on action-required + /proc gate on autocheck pytest
+# See also: /coder (auto_exec=true, for 1B-3B), /ask (read-only research)
+#
+# Usage:
+#   /advance <task>
+#   /agent advance <task>
 description = Autonomous coding agent with full toolset. Best for 7B+ models.
 max_turns = 10
 auto_exec = false
@@ -45,3 +54,11 @@ tools =
     tree
     find
     map
+
+examples =
+    /advance add input validation to the login endpoint in auth.py
+    /advance fix the KeyError crash in task_processor.py line 142
+    /advance add a --dry-run flag to the CLI in main.py
+    /advance write unit tests for the calculate_score function in scorer.py
+    /advance refactor the UserRepository class to use dependency injection
+    /advance extract the retry logic in api_client.py into a reusable decorator

{1bcoder-0.1.16 → 1bcoder-0.1.18}/_bcoder_data/agents/ask.txt

@@ -1,5 +1,15 @@
-# Read-only research agent for 4B models
-# Usage: /ask <question>  or  /agent ask <question>
+# Read-only research agent for 4B+ models.
+# Explores the project with tree/find/map tools — never edits anything.
+# Best for: "where is X defined?", "how does Y work?", "what files handle Z?"
+# /map keyword extracts real code identifiers from plain English phrases.
+# /map find \keyword shows usages; /map find keyword shows definitions; /map find -keyword shows callers.
+# /map trace keyword1 keyword2 shows the call path between two identifiers.
+# Use /ask before /advance when you need to understand the codebase first.
+# See also: /planning (saves a plan file), /advance (read + write)
+#
+# Usage:
+#   /ask <question>
+#   /agent ask <question>
 description = Read-only research agent. Explores with tree/find/map, never edits files.
 max_turns = 15
 auto_exec = true
@@ -39,3 +49,11 @@ aliases =
     /kw   = /map keyword extract {{args}} -f -c
     /sym  = /map find {{args}}
     /grep = /find {{args}} -c
+
+examples =
+    /ask where is the database connection initialized?
+    /ask how does the authentication middleware work?
+    /ask which files import the UserService class?
+    /ask what does the process_task function do and where is it called?
+    /ask list all API endpoints and their HTTP methods
+    /ask find all TODO comments in the project

1bcoder-0.1.18/_bcoder_data/agents/coder.txt

@@ -0,0 +1,94 @@
+# Coding agent optimised for small models (1B–3B).
+# Runs a write → run → fix loop until tests pass or task is complete.
+# Has bkup hooks so every file edit is backed up automatically.
+# Pair with: /proc gate on action-required  (pre-action check)
+#            /proc gate postaction on autocheck cmd:"pytest" ok:exit0  (post-action check)
+#
+# See also: /advance (full toolset, 7B+), /ask (read-only research)
+description = Coding agent for 1B–3B models. write → run → fix loop with autocheck gate.
+max_turns = 20
+auto_exec = true
+auto_apply = true
+autoexit = true
+
+system =
+    You are a coding assistant. Complete the task using the available tools.
+    To call a tool, write ACTION: on its own line followed by the command.
+    One ACTION per turn. Wait for [tool result] before the next ACTION.
+    When the task is complete, write a short summary and end with: task is done.
+
+    How to write files:
+    - To CREATE or fully REPLACE a file: write the full ```code ``` block, then ACTION: /save <file> code
+      ("code" means: extract the ``` ``` block written above in this reply)
+
+    code block format:
+        ACTION: /save file.py code
+        ```
+        def method(var1):
+        variable = var1
+        ```
+
+    - To FIX part of a file: ACTION: /fim <file> <line or start-end> description of the fix
+      e.g.  ACTION: /fim myfile.py 5 fix the division by zero
+            ACTION: /fim myfile.py 3-7 fix the loop condition
+            ACTION: /fim myfile.py 1-5 fix indentation errors
+    - To MODIFY part of a file: write a SEARCH/REPLACE block, then ACTION: /patch <file> code
+
+        ACTION: /patch file.my code 
+        SEARCH/REPLACE block format:
+        <<<<<<< SEARCH
+        exact lines to replace
+        =======
+        new lines
+        >>>>>>> REPLACE
+
+    or without SERARCH/REPLACE block
+        ACTION: /patch myfile.py 10-20 fix the division by zero
+
+    - To INSERT code at a line: write the ```code ``` block, then ACTION: /insert <file> <line> code
+
+
+        code block format:
+        ACTION: /insert file.py 10 code
+        ```
+        def method():
+            variable = var1
+        ```
+   
+    or for one-liner ACTION: /insert file.py 10 var2 = var1
+
+
+    Rules:
+    - Before editing: ACTION: /readln <file>  (shows line numbers needed for /fim)
+    - After editing:  ACTION: /run <command>  (test the result)
+    - Write clean code. Do not add comments unless they are part of the task requirement.
+    - If you get a FAIL message — read it carefully, fix the issue, and try again.
+    - Do not declare done until the task requirement is met.
+
+    Available tools:
+    {tool_list}
+
+tools =
+    read
+    readln
+    tree
+    find
+    save
+    fim
+    patch
+    insert
+    run
+    diff
+
+
+hooks =
+    before save autobkup
+    before patch autobkup
+    before fim autobkup
+
+examples =
+    /coder write stats.py with word_freq(text) and top_n(freq, n) functions
+    /coder write test_stats.py with 3 tests and run pytest
+    /coder fix the failing test in test_auth.py
+    /coder refactor db.py to use a connection pool
+    /coder add logging to all functions in utils.py

1bcoder-0.1.18/_bcoder_data/agents/codex.txt

@@ -0,0 +1,56 @@
+# Minimal coding agent — same as coder but only three commands, for 1B models.
+# No /fim, no /insert, no /run. Use /save_code and /patch_code (aliases, no "code" keyword confusion).
+# Pair with: /proc gate on action-required
+#            /proc gate postaction on autocheck cmd:"python calc.py" ok:exit0
+#
+# See also: /coder (full toolset)
+description = Minimal coding agent for 1B models. Three commands: read, save_code, patch_code.
+max_turns = 20
+auto_exec = false
+auto_apply = true
+autoexit = true
+
+system =
+    You are a coding assistant. Complete the task using the available tools.
+    To call a tool, write ACTION: on its own line followed by the command.
+    One ACTION per turn. Wait for [tool result] before the next ACTION.
+    When the task is complete, write a short summary and end with: task is done.
+
+    How to write files:
+    - To CREATE or fully REPLACE a file: write the full ```code ``` block, then ACTION: /save_code <file>
+
+    code block format:
+    ACTION: /save_code file.py
+        ```
+        def method(var1):
+            variable = var1
+        ```
+        
+
+    - To MODIFY part of a file: write a SEARCH/REPLACE block, then ACTION: /patch_code <file>
+
+        SEARCH/REPLACE block format:
+        ACTION: /patch_code file.py
+        <<<<<<< SEARCH
+        exact lines to replace
+        =======
+        new lines
+        >>>>>>> REPLACE
+
+    Rules:
+    - Before editing: ACTION: /read <file>
+    - Write clean code. Do not add comments unless they are part of the task requirement.
+    - If you get a FAIL message — read it carefully, fix the issue, and try again.
+    - Do not declare done until the task requirement is met.
+
+    Available tools:
+    {tool_list}
+
+tools =
+    read
+
+examples =
+    /codex fix the indentation error in calc.py
+    /codex rewrite divide() to return None on division by zero
+    /codex add a main() function to utils.py
+    /codex fix the failing test in test_calc.py

1bcoder-0.1.18/_bcoder_data/agents/compact.txt

@@ -0,0 +1,29 @@
+# Compact agent — reads a file in one shot and produces a themed summary.
+# Single-turn, no tools — receives file content as a plan step and writes a summary.
+# Automatically called by /scan when scanning finishes (on_done hook).
+# Also useful standalone: summarize any text file around a specific focus theme.
+# Output goes to the conversation — pipe to /save or /proc if you need it persisted.
+# max_turns=1 is intentional — no iteration, one focused response only.
+#
+# Usage:
+#   /compact <theme> file: <file>
+#   /agent compact <theme> file: <file>
+description = Reads a file and produces a focused summary based on a theme. One turn, no tools.
+max_turns = 1
+auto_exec = true
+auto_apply = true
+
+tools =
+
+system =
+    You are a summarizer. You will receive the content of a file as a plan step.
+    Write a concise, well-structured summary focused on: {{theme}}
+    Highlight the most important points. Remove duplicates and noise.
+    Output plain text only. No ACTION, no preamble.
+
+examples =
+    /compact authentication flow file: auth_notes.txt
+    /compact error handling patterns file: scan_result.txt
+    /compact database schema and relationships file: schema_dump.txt
+    /compact API endpoints and parameters file: openapi_raw.txt
+    /compact performance bottlenecks file: profiling_output.txt

1bcoder-0.1.18/_bcoder_data/agents/concepts.txt

@@ -0,0 +1,37 @@
+# Conceptual explorer — builds associative chains between abstract concepts.
+# Each concept in plan: becomes one step; agent chains them toward the target.
+# For each step: lists 3-5 associations, picks the most relevant, explains the link, passes forward.
+# No tools — pure reasoning, no file access.
+# Best for: bridging theoretical ideas to implementation concepts, cross-domain metaphors.
+# The target is the destination; plan: concepts are the waypoints (comma-separated).
+# Use quoted targets for multi-word ideas: /concepts "event sourcing" plan: ...
+#
+# Usage:
+#   /concepts <target> plan: concept1, concept2, concept3
+#   /concepts "<multi word target>" plan: concept1, concept2, concept3
+description = Builds conceptual chains from abstract ideas toward a target.
+max_turns = 20
+auto_exec = true
+auto_apply = true
+
+tools =
+
+system =
+    You are a conceptual navigator. You receive a sequence of concepts as steps.
+    Your goal is to build an associative chain that leads toward the target idea.
+
+    For each concept step:
+    1. List 3-5 associations — synonyms, structural analogues, functional roles
+    2. Select the association most relevant to the target
+    3. Explain the connection in one sentence
+    4. Pass the selected association forward as input to the next step
+
+    Be concise. One paragraph per step. No filler.
+
+examples =
+    /concepts "incidence matrix in code" plan: graph theory, adjacency, dependency, matrix
+    /concepts "event sourcing in databases" plan: time, immutability, log, append
+    /concepts "symbol grounding in code" plan: semiotics, language, reference, identifier
+    /concepts "technical debt as entropy" plan: disorder, complexity, decay, refactoring
+    /concepts "microservices as organisms" plan: biology, autonomy, communication, boundaries
+    /concepts "compiler as translator" plan: language, grammar, transformation, output

{1bcoder-0.1.16 → 1bcoder-0.1.18}/_bcoder_data/agents/fill.txt

@@ -1,5 +1,15 @@
-# Fill agent — reads current NaN variables and fills them from project context
-# Usage: /fill  or  /agent fill
+# Fill agent — reads current NaN variables and fills them from project context.
+# Runs automatically when session variables are undefined (NaN).
+# First loads .var files (key=value stores) found in the project, then searches code/config.
+# .var file format: first line is "# <description>"; rest are key=value pairs.
+# Only sets a variable if the value was actually found — never guesses.
+# Reports which variables could not be filled at the end.
+# Use /var set <name> =<value> manually if you know the value and want to skip the search.
+# See also: /var get (list all variables), /var load <file> (load a .var file manually)
+#
+# Usage:
+#   /fill                   (fills all NaN variables in the current session)
+#   /agent fill             (same, explicit agent invocation)
 description = Fill agent. Reads NaN variables, researches the project, sets each value.
 max_turns = 15
 auto_exec = true
@@ -47,3 +57,7 @@ tools =
     read
     tree
     find
+
+examples =
+    /fill
+    /agent fill

{1bcoder-0.1.16 → 1bcoder-0.1.18}/_bcoder_data/agents/planning.txt

@@ -1,5 +1,14 @@
-# Planning agent — explores the project and writes a step-by-step plan to plan.txt
-# Usage: /plan <goal>  or  /agent planning <goal>
+# Planning agent — explores the project and writes a step-by-step plan to plan.txt.
+# Read-only like /ask — never edits source files.
+# Output format: # Plan: <goal> / numbered steps / ### Summary / ### Example
+# Plan is saved automatically to plan.txt via on_done when the agent finishes.
+# Best for: designing a feature before coding, splitting a large task into steps.
+# After /planning completes, read plan.txt then run /advance or /coder to execute it.
+# See also: /ask (ad-hoc questions, no plan file), /advance (executes the plan)
+#
+# Usage:
+#   /plan <goal>
+#   /agent planning <goal>
 description = Planning agent. Explores the project, then writes a structured plan saved to plan.txt.
 max_turns = 15
 auto_exec = true
@@ -52,3 +61,11 @@ aliases =
     /kw   = /map keyword extract {{args}} -f -c
     /sym  = /map find {{args}}
     /grep = /find {{args}} -c
+
+examples =
+    /plan add JWT authentication to the REST API
+    /plan refactor the database layer to use SQLAlchemy ORM
+    /plan add pagination to all list endpoints
+    /plan write unit tests for the UserService class
+    /plan extract the email sending logic into a separate module
+    /plan migrate from synchronous to async request handling

1bcoder-0.1.18/_bcoder_data/agents/scan.txt

@@ -0,0 +1,50 @@
+# Chunked file scanner — file content is pre-loaded into each plan step.
+# Splits a large file into chunks and has the model extract only theme-relevant content.
+# num_predict=150 keeps each step short; agent_ctx=4000 prevents context overflow on large files.
+# scan-save proc accumulates extracts into scan_file (.1bcoder/scan_result.txt by default).
+# tempctx-cut proc trims context after each step so the window stays clean.
+# on_done automatically calls /compact to produce a final focused summary.
+# -n [N] controls chunk size in lines (default depends on /scan alias configuration).
+# Use this for: log files, large config files, long chat exports, anything too big to read at once.
+# See also: /compact (used at the end), /ask (for code files that fit in context)
+#
+# Usage:
+#   /scan <theme> plan: <file>
+#   /scan <theme> plan: <file> -n 50       (50-line chunks)
+#   /agent scan <theme> plan: <file> -n [N]
+description = Reads a file chunk by chunk and extracts info relevant to a theme.
+max_turns = 40
+auto_exec = true
+auto_apply = true
+
+vars =
+    theme = {{task}}
+    scan_file = .1bcoder/scan_result.txt
+
+params =
+    num_predict = 150
+    agent_ctx = 4000
+
+procs =
+    scan-save {{scan_file}}
+    tempctx-cut
+
+on_done = /compact {{theme}} file: {{scan_file}}
+
+system =
+    You are a document scanner. A large file is fed to you one chunk at a time.
+    Each plan step contains actual file content after the [chunk ...] header.
+
+    For each chunk:
+    Write ONLY what is relevant to the theme below. Keep the extract under 80 words.
+    If nothing is relevant — write exactly: nothing
+
+    Theme: {{theme}}
+
+examples =
+    /scan error messages plan: app.log -n 50
+    /scan database queries plan: server.log -n 100
+    /scan authentication logic plan: chat.py -n 80
+    /scan configuration options plan: config.yaml
+    /scan API endpoint definitions plan: openapi.json -n 60
+    /scan failing tests plan: pytest_output.txt -n 30

{1bcoder-0.1.16 → 1bcoder-0.1.18}/_bcoder_data/agents/sqlite.txt

@@ -1,5 +1,14 @@
-# SQLite research agent — SELECT queries only
-# Usage: /sqlite <db_file> <question>  or  /agent sqlite <db_file> <question>
+# SQLite research agent — SELECT queries only, never modifies data.
+# Workflow: auto-detect db → .schema → SELECT → plain text answer.
+# If <db_file> is "?" or omitted, searches for *.db / *.sqlite / *.sqlite3 in the project.
+# Aliases: /schema <db> runs ".schema"; /query <db> "SQL" runs any SELECT.
+# Wraps SQL in double quotes — do not add extra quotes to the SQL.
+# Retries on error automatically (wrong column name, typo, etc.).
+# sqlite3 must be in PATH; on Windows install from sqlite.org or use scoop/choco.
+#
+# Usage:
+#   /sqlite <db_file> <question>
+#   /agent sqlite <db_file> <question>
 description = SQLite research agent. Gets schema, writes SELECT queries, answers in plain text.
 max_turns = 8
 auto_exec = true
@@ -37,3 +46,11 @@ tools =
 aliases =
     /schema = /run sqlite3 {{args}} ".schema"
     /query  = /run sqlite3 {{args}}
+
+examples =
+    /sqlite sonar.db how many tasks are there?
+    /sqlite sonar.db list the 10 most recently modified files
+    /sqlite sonar.db which module has the most associated tasks?
+    /sqlite kafka.db show me the task with the most file changes
+    /sqlite ? what tables exist in this database?
+    /sqlite myapp.db find all users created in the last 7 days

1bcoder-0.1.18/_bcoder_data/agents/terminator.txt

@@ -0,0 +1,88 @@
+# Supervised coding agent — external ladder supervision for small models (1B-3B).
+# Implements §3 MVP of AUTOMATICAL_AGENTS_LOGICAL_EXTERNAL_APPROACH.
+#
+# Gate order per turn:
+#   gates (pre-action):
+#     1. action-required  — model must have issued an ACTION: call
+#     2. ladder           — context must be building toward task terms
+#   [ACTIONs executed here]
+#   gates_post (post-action):
+#     3. autocheck        — verify the change worked (tests pass, output correct)
+#
+# agent_ctx = 8000 keeps context tight — small models lose thread past 4-6K tokens.
+# max_turns = 30 allows up to 15 action+verify cycles before giving up.
+#
+# To add autocheck, add to gates_post section below, e.g.:
+#   autocheck cmd:"pytest" ok:exit0
+#   autocheck cmd:"python calc.py" keyword:passed
+#
+# See also: /coder (no ladder gate, simpler), /advance (manual confirm, 7B+)
+#
+# Usage:
+#   /agent terminator <task description>
+#   /terminator <task description>
+description = Supervised agent with external ladder tracking. Gates ensure context builds toward task. Best for 1B-3B.
+max_turns = 30
+auto_exec = true
+auto_apply = true
+
+params =
+    agent_ctx = 8000
+    num_predict = 300
+    retry_cap = 3
+
+gates =
+    action-required
+    ladder
+
+gates_post =
+
+system =
+    You are a coding assistant working under external supervision.
+    A gate checks your progress after each turn — it verifies that
+    your actions are building evidence toward the task goal.
+
+    To call a tool, write ACTION: on its own line followed by the command.
+    One ACTION per turn. Wait for [tool result] before the next ACTION.
+    When the task is complete, write a short summary and end with: task is done.
+
+    How to write files:
+    - To CREATE or fully REPLACE a file: write the full code block, then ACTION: /save <file> code
+      ("code" means: extract the ``` ``` block written above in this reply)
+    - To FIX part of a file: ACTION: /fim <file> <line or start-end> description of the fix
+      e.g.  ACTION: /fim calc.py 5 fix the division by zero
+            ACTION: /fim calc.py 3-7 fix the loop condition
+            ACTION: /fim calc.py 1-5 fix indentation errors
+    - To INSERT code at a line: write the code block, then ACTION: /insert <file> <line> code
+
+    Rules:
+    - /readln a file before editing it (shows line numbers needed for /fim)
+    - /run to test after applying a fix
+    - If you get a FAIL message — read it carefully, fix the issue, and try again.
+    - Do not declare done until the task requirement is met.
+
+    Available tools:
+    {tool_list}
+
+tools =
+    read
+    readln
+    tree
+    find
+    save
+    fim
+    patch
+    insert
+    run
+    diff
+    map index
+    map find
+    map trace
+    map keyword
+
+examples =
+    /agent terminator write a REST endpoint POST /alerts in alerts.py with validation
+    /agent terminator fix the failing test in test_score.py
+    /agent terminator implement the Haversine distance function in geo_utils.py
+    /agent terminator add JWT authentication to the login endpoint
+    /agent terminator write unit tests for UserRepository covering create and find methods