llms.txt
Compare with other sites' llms.txt files →logcat.ai - The AI engineering layer for the OS stack
===============================================================================
Full reference: https://logcat.ai/llms-full.txt
(long form: every supported file class, what each one answers, the
investigation modes, pricing, and common questions, inlined)
Description:
------------
logcat.ai is an AI-native platform for Android and Linux device systems
engineering. It takes the artifacts a device actually produces, bugreports,
logcat, kernel logs, traces, packet captures, memory dumps and robot recordings,
and turns them into grounded findings an engineer can act on: what failed, where,
and what the evidence for that is.
Every finding cites the log lines it came from. The model proposes; the engineer
decides.
Analyzers:
----------
Eight file classes are analyzed, each with its own dashboard and its own view of
the device.
1. BUGREPORT ANALYZER (/bugreport-analyzer)
• Full Android bugreport archives, including 100MB+ captures from adb,
Android Studio, or CI
• Parallel subsystem analysis covering applications, system services, memory,
power, network, storage, security, CPU and performance, timeline, and
overall system health
• Cross-subsystem correlation, so a battery symptom can be traced to the
service actually responsible
• Embedded logcat and kernel viewers alongside the findings
2. LOGCAT ANALYZER (/logcat-analyzer)
• Android application and system logs from adb, Android Studio, or CI output
• Crash, ANR, and lifecycle detection with severity ranking
• Timeline view of activity spikes and error clustering
• Filtering by tag, priority, PID, and time range
3. DMESG / KERNEL ANALYZER (/dmesg-analyzer)
• Linux kernel logs from any distribution or embedded target
• Panic, oops, warning, and driver failure detection
• Works across ARM, arm64, x86, and RISC-V targets
• Subsystem attribution for block I/O, networking, memory management, and
scheduling
4. PERFETTO TRACE ANALYZER (/perfetto-analyzer)
• Perfetto and systrace captures queried as structured data rather than
flattened to text
• Scheduling, frame timing, and contention analysis
5. PCAP ANALYZER (/pcap-analyzer)
• pcap and pcapng captures, decoded automatically on upload
• Flow, protocol, and failure analysis over the real packet records
• Capture with tcpdump or Wireshark and upload the file directly
6. BUNDLE ANALYZER (/bundle-analyzer)
• Multi-file archives and OEM log bundles
• Files are related to each other, so an identifier in one file resolves
against the others in the same bundle
• Per-file dashboards plus a bundle-level view
7. RAMDUMP ANALYZER (/ramdump-analyzer)
• Kernel ramdumps, recognized by their contents rather than by archive type
• Reset reason, interconnect faults, per-CPU run queues, task and memory
state, and recovered device logs
• Upload a matching vmlinux to the reference library to resolve kernel
symbols and go deeper on later dumps of the same build
8. ROSBAG ANALYZER (/rosbag-analyzer)
• ROS recordings in all three formats, recognized by their contents rather
than by file name
• Channel inventory with message types and counts, the recording's own time
window, and each component's self-reported health
• The robot's log stream is pulled out into a searchable log beside the
recording, so its own account of the run is readable and queryable
• Sensor payload is deliberately left undecoded, which is what makes a very
large recording quick to analyze
Investigation modes:
--------------------
QUICK SEARCH
• A single-pass question against an uploaded file
• Natural language in, cited answer out, typically in seconds
• Suited to triage and to locating something you already suspect
DEEP RESEARCH (/platform/diagnose)
• Multi-step investigation that plans its own queries, follows evidence
across subsystems, and synthesizes a cited report
• Suited to open questions where the failing component is not yet known
• Progress is visible while it runs, and the report is shareable
DELTA
• Structured comparison across several files or captures
• Answers what changed between a working and a failing device, build, or run
Solutions:
----------
• Device manufacturers (/solutions/device-manufacturers) - cross-layer
root cause for OEMs and ODMs across bugreports, Perfetto traces, modem
captures, kernel logs, memory dumps and their own system-app logs
• Automotive (/solutions/automotive) - vehicle software across the stack,
including CAN traces with an uploaded DBC for decoding
• Modem and connectivity (/solutions/modem) - baseband, RIL, and IMS logs
• Embedded Linux and silicon (/solutions/embedded-linux-silicon) - Yocto, BSP,
and bring-up work
• Robotics (/solutions/robotics) - ROS recordings read alongside the kernel
logs, captures and traces from the same robot
• Forensics and audit (/solutions/forensics) - device management state,
enforced policy, traffic analysis and two-point comparison
• Wi-Fi and access points (/solutions/wifi) - access point logs, router config and
client fingerprint tables read together; clients resolved to named devices
• CTS and VTS testing (/solutions/cts-vts-testing) - compliance test failure
analysis
Platform:
---------
• How it works (/platform/architecture) - architecture, how findings are
grounded in the source logs, and the security model
• Diagnose (/platform/diagnose) - the investigation engine behind Deep Research
Pricing:
--------
• Free tier for individual engineers
• Enterprise plans for teams, with capability-based access to the telecom,
automotive, ramdump and robotics verticals
• Usage is metered in investigation units, so a quick question costs a fraction
of a deep investigation
• Details at https://logcat.ai/pricing
Open source program:
--------------------
• Free access for open source maintainers and contributors
• Details at https://logcat.ai/oss
Upload formats:
---------------
• Android bugreport archives and text dumps
• logcat in the common formats, including threadtime and brief
• Linux kernel logs and dmesg output
• Perfetto and systrace captures
• pcap and pcapng packet captures
• Kernel ramdumps and OEM log bundles
• Automotive traces, including DLT and CAN
• ROS robot recordings, in all three recording formats
Who it is for:
--------------
• Android platform and application engineers
• Linux kernel and driver developers
• OEM and ODM platform teams
• Embedded and BSP engineers
• Modem and connectivity engineers
• Automotive software teams
• Robotics platform and field autonomy engineers
• QA, test automation, and support engineers
Getting started:
----------------
1. Sign up at https://logcat.ai
2. Upload a bugreport, logcat, kernel log, trace, capture, or dump
3. Read the findings, each linked to the log lines behind it
4. Ask a follow-up question with Quick Search
5. Run Deep Research when the failing component is not yet known
Links:
------
• Product: https://logcat.ai
• Documentation: https://docs.logcat.ai
• Blog: https://blog.logcat.ai
• Pricing: https://logcat.ai/pricing
• Security and architecture: https://logcat.ai/platform/architecture
Contact:
--------
• Sales and commercial enquiries: sales@logcat.ai
• Security reports: security@logcat.ai
Privacy and security:
---------------------
Data handling, retention, and processing are described in the privacy policy at
https://logcat.ai/privacy. The security model and how uploaded data is handled
are described at https://logcat.ai/platform/architecture.
---
(c) 2026 Log Operations And Genai Enterprises, Inc. All Rights Reserved.