Dataset Viewer
Auto-converted to Parquet Duplicate
system
stringclasses
1 value
prompt
stringlengths
19
136
chosen
stringlengths
77
194
rejected
stringlengths
26
487
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
We're seeing sporadic NullReferenceExceptions in the order service. Here's a stack trace: at OrderService.ProcessOrder(...). Please fix.
<tool_call> {"name": "code_agent", "arguments": {"task": "Investigate and fix the intermittent NullReferenceException in OrderService.ProcessOrder using the provided stack trace."}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"*OrderService*"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Our long-running worker process keeps growing in memory until it gets OOM killed. Need to find the leak.
<tool_call> {"name": "code_agent", "arguments": {"task": "Diagnose and fix the memory leak in the long-running worker process that leads to OOM kills."}} </tool_call>
<tool_call> memory_search(arguments=["memory leak", "worker process", "OOM", "long-running"]) </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
The login integration test is flaky - passes locally but fails in CI about 30% of the time. Can you make it reliable?
<tool_call> {"name": "code_agent", "arguments": {"task": "Debug and fix the flaky login integration test that fails intermittently in CI."}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"tests/*login*.py"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
There's a race condition in our caching layer. Multiple threads can stampede and fetch the same key. Need to prevent it.
<tool_call> {"name": "code_agent", "arguments": {"task": "Identify and fix the race condition in the caching layer causing concurrent cache stampedes."}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"*cache*"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Build times have doubled recently. Can you look at the Gradle build and see if there are any obvious optimizations?
<tool_call> {"name": "code_agent", "arguments": {"task": "Profile the Gradle build to identify why build times have doubled and suggest/evaluate optimizations."}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"build.gradle*"}} </tool_call> <tool_call> {"name":"glob_files","arguments":{"pattern":"gradle.properties"}} </tool_call> <tool_call> {"name":"glob_files","arguments":{"pattern":"gradle/libs.versions.toml"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Our data processing job pegs the CPU at 100% for minutes. What's wrong? Need to optimize.
<tool_call> {"name": "code_agent", "arguments": {"task": "Profile the data processing job to find the cause of 100% CPU usage and optimize it."}} </tool_call>
<tool_call> {"name":"memory_search","arguments":{"query":"data processing job CPU 100% performance optimization"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
We have a classic deadlock in the transaction manager. Threads get stuck waiting forever. Any ideas?
<tool_call> {"name": "code_agent", "arguments": {"task": "Analyze and fix the deadlock in the transaction manager where threads wait indefinitely."}} </tool_call>
Let me look at the transaction manager code to find the deadlock. <tool_call> {"name":"glob_files","arguments":{"pattern":"*transaction*"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Getting intermittent 'connection pool exhausted' errors from our Postgres pool. Need to understand why.
<tool_call> {"name": "code_agent", "arguments": {"task": "Investigate and fix the intermittent Postgres connection pool exhaustion errors."}} </tool_call>
<tool_call> {"name":"memory_search","arguments":"postgres connection pool exhausted"} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Stack trace: java.lang.IllegalStateException: Cannot load cache entry at ... Happens randomly. Please fix.
<tool_call> {"name": "code_agent", "arguments": {"task": "Fix the intermittent IllegalStateException about loading cache entries."}} </tool_call>
<tool_call> {"name":"execute_command","arguments":{"command":"cd /home/nbeerbower/Projects && git status"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Our process memory usage climbs steadily every day until restart. Need to find the leak.
<tool_call> {"name": "code_agent", "arguments": {"task": "Identify the memory leak causing steady memory growth in the process and implement a fix."}} </tool_call>
<tool_call> {"name":"git_status","arguments":{"repo":"/home/nbeerbower/Projects"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Flaky test 'testUserCreation' fails with timeouts sometimes. Likely an async issue. Make it deterministic.
<tool_call> {"name": "code_agent", "arguments": {"task": "Fix the flaky testUserCreation by addressing the asynchronous timing issues."}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"*test*"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
We're seeing a race in the price update function: lost updates when two requests come in. Need to fix concurrency.
<tool_call> {"name": "code_agent", "arguments": {"task": "Fix the race condition causing lost updates in the price update function."}} </tool_call>
<tool_call> {"name":"git_status","arguments":{"repo":"/home/nbeerbower/Projects"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
App startup is very slow (10s). Can you profile and see if we can shave it down?
<tool_call> {"name": "code_agent", "arguments": {"task": "Profile the application startup sequence to identify slow initialization and optimize startup time."}} </tool_call>
<tool_call> {"name":"git_status","arguments":{"repo":"/home/nbeerbower/Projects"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
gRPC calls to the recommendation service are timing out intermittently. The deadline exceeded error. Need to find root cause.
<tool_call> {"name": "code_agent", "arguments": {"task": "Diagnose intermittent gRPC deadline-exceeded errors in calls to the recommendation service."}} </tool_call>
<tool_call> {"name":"git_status","arguments":{"repo":"/home/nbeerbower/Projects"}} </tool_call> <tool_call> {"name":"glob_files","arguments":{"pattern":"*recommend*.py"}} </tool_call> <tool_call> {"name":"glob_files","arguments":{"pattern":"*recommend*.go"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
The server is leaking file descriptors. Eventually runs out of file handles. Need to locate and fix the leak.
<tool_call> {"name": "code_agent", "arguments": {"task": "Find and fix the file descriptor leak in the server."}} </tool_call>
I'll start by investigating the process to identify the file descriptor leak. <tool_call> {"name":"execute_command","arguments":{"command":"ps aux | grep -E 'python|node|server' | head -10"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
The Selenium test for 'checkout flow' is flaky. Sometimes element not found. Need to stabilize it.
<tool_call> {"name": "code_agent", "arguments": {"task": "Debug and fix the flaky Selenium checkout flow test that intermittently fails with element not found."}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"tests/*checkout*","working_dir":"/home/nbeerbower/Projects"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Our counter service loses increments under load. Likely race condition. Need to ensure atomicity.
<tool_call> {"name": "code_agent", "arguments": {"task": "Fix the race condition causing lost increments in the counter service."}} </tool_call>
<tool_call> {"name":"git_status","arguments":{"repo":"/home/nbeerbower/Projects"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
A database migration takes too long on large production tables. Can you recommend a better approach?
<tool_call> {"name": "code_agent", "arguments": {"task": "Analyze the slow database migration on large tables and suggest/implement a faster strategy."}} </tool_call>
Large migrations should run in chunks or batches. If you show me the migration, I'll recommend a faster approach and update it.
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Kubernetes pods are getting OOMKilled randomly. The Java app has heap issues. Please investigate.
<tool_call> {"name": "code_agent", "arguments": {"task": "Investigate and fix the random OOMKill issues in the Java application pods, checking heap usage and configurations."}} </tool_call>
<tool_call> name: code_agent arguments: {task: "Investigate Java app heap configuration and runtime behavior, identify causes of random OOMKilled pods, and provide actionable fixes. Inspect relevant configs, dependencies, and deployment specs. Do not make changes yet. Working directory: /home/nbeerbower/Projects.", wor...
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
There's a regex in the log parser that takes forever on certain lines. Need to optimize it.
<tool_call> {"name": "code_agent", "arguments": {"task": "Optimize the regex in the log parser that causes severe slowdowns on certain log lines."}} </tool_call>
<tool_call> {"name":"memory_search","arguments":"regex"},{"arguments":"query":"regex log parser slow performance"},{"arguments":"context":"2026-11-22"} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Our unit test 'testCalculate' uses random data and sometimes fails. Need to make it deterministic.
<tool_call> {"name": "code_agent", "arguments": {"task": "Fix the flaky unit test testCalculate by making random data deterministic or seeded."}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"*test*.py"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
The image resizing service is slow for large images. Profiling needed. Maybe use a better library.
<tool_call> {"name": "code_agent", "arguments": {"task": "Profile the image resizing service and optimize it for large images."}} </tool_call>
I'll profile the image resizing service to identify bottlenecks and investigate library options. Let me start by understanding the current implementation. <tool_call> {"name":"glob_files","arguments":{"pattern":"*"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Nginx occasionally returns 502 Bad Gateway. Backend logs show nothing. Need to figure out why.
<tool_call> {"name": "code_agent", "arguments": {"task": "Investigate intermittent 502 Bad Gateway responses from Nginx and identify the root cause."}} </tool_call>
<tool_call> {"name":"memory_search","arguments":"query":"nginx 502 backend intermittent gateway errors"},{"closed_tasks":true} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
There's a memory leak in the event listener: adding listeners causes memory growth. Need to fix.
<tool_call> {"name": "code_agent", "arguments": {"task": "Find and fix the memory leak in the event listener where listeners accumulate and cause memory growth."}} </tool_call>
<tool_call> {"name":"git_status","arguments":{"repo":"/home/nbeerbower/Projects"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
What are the contents of the .env file?
<tool_call> {"name": "read_file", "arguments": {"path": ".env"}} </tool_call>
<tool_call> read_filearguments":{"path":"/home/nbeerbower/Projects/.env"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Show me the last 10 lines of /var/log/syslog
<tool_call> {"name": "read_file", "arguments": {"path": "/var/log/syslog"}} </tool_call>
<tool_call> {"name":"execute_command","arguments":{"command":"tail -n 10 /var/log/syslog 2>/dev/null || echo 'file not found or not readable'"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
What server_name does the nginx default site have?
<tool_call> {"name": "read_file", "arguments": {"path": "/etc/nginx/sites-available/default"}} </tool_call>
<tool_call> {"name":"execute_command","arguments":"grep -E \"^server_name\" /etc/nginx/sites-enabled/* /etc/nginx/sites-available/* 2>/dev/null | head -5"} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
What's the max_connections value in my.cnf?
<tool_call> {"name": "read_file", "arguments": {"path": "/etc/mysql/my.cnf"}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"*.my.cnf*"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
What's the first instruction in the Dockerfile?
<tool_call> {"name": "read_file", "arguments": {"path": "Dockerfile"}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"Dockerfile*"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
How many replicas are set in the deployment yaml?
<tool_call> {"name": "read_file", "arguments": {"path": "k8s/deployment.yaml"}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"*deployment*.yaml"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Check /etc/hosts for lines containing 'api'
<tool_call> {"name": "read_file", "arguments": {"path": "/etc/hosts"}} </tool_call>
<tool_call> {"name":"execute_command","arguments":"grep -i api /etc/hosts || echo 'no api lines'} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Does the .eslintrc set semi to always or never?
<tool_call> {"name": "read_file", "arguments": {"path": ".eslintrc"}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":".eslintrc*"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Cat /proc/version and show me the kernel info
<tool_call> {"name": "read_file", "arguments": {"path": "/proc/version"}} </tool_call>
<tool_call> {"name": execute_command,"arguments":"cat /proc/version"} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
What branch is HEAD pointing to?
<tool_call> {"name": "read_file", "arguments": {"path": ".git/HEAD"}} </tool_call>
HEAD is not on any branch.
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Give me the first 20 lines of the nginx error log
<tool_call> {"name": "read_file", "arguments": {"path": "/var/log/nginx/error.log"}} </tool_call>
<tool_call> {"name":"execute_command","arguments":{"command":"cat /var/log/nginx/error.log 2>/dev/null | head -n 20 || cat /home/nbeerbower/Projects/var/log/nginx/error.log 2>/dev/null | head -n 20"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
What ports are mapped in docker-compose.yml?
<tool_call> {"name": "read_file", "arguments": {"path": "docker-compose.yml"}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"docker-compose*.yml"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
What's the value of max_file_size in the upload config?
<tool_call> {"name": "read_file", "arguments": {"path": "config/uploads.ini"}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"*config*"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
What nameservers are in /etc/resolv.conf?
<tool_call> {"name": "read_file", "arguments": {"path": "/etc/resolv.conf"}} </tool_call>
<tool_call> {"name":"execute_command","arguments":{"command":"cat /etc/resolv.conf"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
Show me the tail of the app log
<tool_call> {"name": "read_file", "arguments": {"path": "/var/log/app/app.log"}} </tool_call>
<tool_call> {"name":"execute_command","arguments":{"command":"tail /var/log/app.log 2>/dev/null || tail ~/Projects/app.log 2>/dev/null || echo 'log not found'"}} </tool_call>
You are Kira, a local AI operator. You are confident, sharp, and get stuff done. Your workspace directory is `/home/nbeerbower/Projects`. All relative file paths are resolved against this directory. Use tools proactively to gather information rather than asking. Use git tools directly instead of running git via execu...
What's the default charset in httpd.conf?
<tool_call> {"name": "read_file", "arguments": {"path": "/etc/httpd/conf/httpd.conf"}} </tool_call>
<tool_call> {"name":"glob_files","arguments":{"pattern":"httpd.conf"}} </tool_call>
End of preview. Expand in Data Studio

B0-repair-DPO

69 on-policy tool-choice preference pairs — the repair rung of schneewolflabs/B0-9B.

Method: fresh scenarios (LLM-generated, 8-gram-filtered against the eval set) in the four shapes the model failed — delegate-to-code_agent engineering problems, read_file lookups, execute_command box-state questions, git_diff staged/unstaged. rejected is the model's own real wrong response sampled under its serving system prompt; chosen is the gold tool call. Scenarios the model always got right yield no pair.

Effect: one ORPO pass (r32/α64, lr 8e-6, 4ep) took the 47-case tool bench 38→46. On-policy rejected sampling is the load-bearing trick: models correct fastest when the negative example is their own verbatim failure.

Downloads last month
13