mkdocs.yml

site_name: LLMTrace
site_url: https://docs.llmtrace.io
site_description: Zero-code LLM observability and security for production. Transparent proxy that captures, analyses, and secures LLM interactions in real-time.
site_author: LLMTrace contributors
repo_url: https://github.com/epappas/llmtrace
repo_name: epappas/llmtrace
edit_uri: edit/main/docs/
copyright: Copyright © LLMTrace contributors. Licensed under MIT.

theme:
  name: material
  custom_dir: _overrides
  language: en
  features:
    - navigation.sections
    - navigation.expand
    - navigation.top
    - navigation.tabs
    - navigation.tabs.sticky
    - navigation.instant
    - navigation.tracking
    - navigation.footer
    - toc.follow
    - search.suggest
    - search.highlight
    - search.share
    - content.code.copy
    - content.code.annotate
    - content.tabs.link
    - content.action.edit
    - content.action.view
  palette:
    - media: "(prefers-color-scheme: light)"
      scheme: default
      primary: indigo
      accent: indigo
      toggle:
        icon: material/brightness-7
        name: Switch to dark mode
    - media: "(prefers-color-scheme: dark)"
      scheme: slate
      primary: indigo
      accent: indigo
      toggle:
        icon: material/brightness-4
        name: Switch to light mode
  icon:
    repo: fontawesome/brands/github

extra:
  social:
    - icon: fontawesome/brands/github
      link: https://github.com/epappas/llmtrace
    - icon: fontawesome/brands/rust
      link: https://crates.io/crates/llmtrace
    - icon: fontawesome/brands/python
      link: https://pypi.org/project/llmtracing/

markdown_extensions:
  - abbr
  - admonition
  - attr_list
  - def_list
  - footnotes
  - md_in_html
  - tables
  - toc:
      permalink: true
  - pymdownx.highlight:
      anchor_linenums: true
      line_spans: __span
      pygments_lang_class: true
  - pymdownx.inlinehilite
  - pymdownx.snippets:
      base_path:
        - .
        - docs
      check_paths: true
  - pymdownx.superfences:
      custom_fences:
        - name: mermaid
          class: mermaid
          format: !!python/name:pymdownx.superfences.fence_code_format
  - pymdownx.tabbed:
      alternate_style: true
  - pymdownx.tasklist:
      custom_checkbox: true
  - pymdownx.details
  - pymdownx.caret
  - pymdownx.mark
  - pymdownx.tilde
  - pymdownx.keys
  - pymdownx.smartsymbols
  - pymdownx.emoji:
      emoji_index: !!python/name:material.extensions.emoji.twemoji
      emoji_generator: !!python/name:material.extensions.emoji.to_svg

plugins:
  - search
  - timetoread:
      wpm: 210
  - table-reader
  - tags
  - minify:
      minify_html: true
      minify_css: true

hooks:
  - hooks/llm_utils.py

extra_javascript:
  - javascripts/mathjax.js
  - https://polyfill.io/v3/polyfill.min.js?features=es6
  - https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js
  - https://unpkg.com/tablesort@5.3.0/dist/tablesort.min.js
  - javascripts/tablesort.js
  - javascripts/tts.js
  - javascripts/copy-llm.js

extra_css:
  - css/custom.css

exclude_docs: |
  README.md
  prompts/
  TODO.md
  TODO_E2E.md

validation:
  omitted_files: warn
  absolute_links: info
  unrecognized_links: info
  anchors: info
  nav:
    omitted_files: info
    not_found: warn
  links:
    not_found: warn
    absolute_links: info
    anchors: info
    unrecognized_links: info

nav:
  - Home: index.md
  - Getting Started:
      - Installation: getting-started/installation.md
      - Quickstart: getting-started/quickstart.md
      - Configuration: getting-started/configuration.md
  - Guides:
      - OpenAI SDK: guides/integration-openai.md
      - LangChain: guides/integration-langchain.md
      - curl / HTTP: guides/integration-curl.md
      - Custom Policies: guides/custom-policies.md
      - Pre-Request Enforcement: guides/enforcement.md
      - LLM Judge Setup: guides/llm-judge.md
      - Auth & Multi-Tenancy: guides/auth-tenancy.md
      - Dashboard: guides/dashboard.md
      - Monitoring: guides/monitoring.md
      - Troubleshooting: guides/troubleshooting.md
      - Integration Tests: guides/integration-tests.md
      - E2E Adversarial Testing: guides/e2e-testing.md
      - Releasing: guides/releasing.md
      - REST API: guides/API.md
  - Security & ML:
      - Ensemble Detection: ml/ensemble.md
      - Model Reference: ml/models.md
      - Threshold Tuning: ml/tuning.md
      - Benchmark Methodology: ml/benchmarks.md
      - OWASP LLM Top 10: security/OWASP_LLM_TOP10.md
      - OWASP GenAI Architecture: security/OWASP_GENAI_TOP10_2025_ARCHITECTURE.md
      - Analysers Breakdown: security/ANALYSERS_TECHNICAL_BREAKDOWN.md
  - Deployment:
      - Docker Compose: deployment/docker-compose.md
      - Kubernetes: deployment/kubernetes.md
      - Secrets Management: deployment/secrets-management.md
  - Operations:
      - Feature Flags: runbooks/feature-flags.md
      - Judge Golden-Set Drift: runbooks/judge-golden-set-drift.md
      - Dashboard Features: dashboard-features.md
  - Architecture:
      - System Architecture: architecture/SYSTEM_ARCHITECTURE.md
      - Transparent Proxy: architecture/TRANSPARENT_PROXY.md
      - LLM Judge: architecture/LLM_JUDGE.md
      - Judge Cascade (ADR): architecture/JUDGE_CASCADE.md
      - Boundary Token Defence: architecture/BOUNDARY_TOKEN_DEFENCE.md
      - Spotlighting Indirect Injection (Investigation): architecture/SPOTLIGHTING_INDIRECT_INJECTION.md
      - ML Long-Input Defence: architecture/ML_LONG_INPUT_DEFENCE.md
      - DMPI Average Pooling: architecture/DMPI_001_AVERAGE_POOLING.md
      - DMPI Two FC Layers: architecture/DMPI_002_TWO_FC_LAYERS.md
      - DMPI Ten Binary Features: architecture/DMPI_003_TEN_BINARY_FEATURES.md
  - Research:
      - Overview: research/index.md
      - Landscape & OWASP:
          - Landscape: research/landscape.md
          - State of the Art 2026: research/security-state-of-art-2026.md
          - Benchmarks & Tools Landscape: research/benchmarks-and-tools-landscape.md
          - OWASP GenAI Top 10 2025 References: research/owasp-genai-top10-2025-references.md
          - LLMTrace OWASP GenAI Architecture Report: research/llmtrace-owasp-genai-architecture-report.md
          - LLMTrace Defence Pipeline Design: research/llmtrace-defence-pipeline-design.md
      - Attacks & Benchmarks:
          - BIPIA Indirect Prompt Injection: research/bipia-indirect-prompt-injection-benchmark.md
          - CyberSecEval2: research/cyberseceval2-llm-security-benchmark.md
          - WASP Web-Agent Security: research/wasp-web-agent-security-benchmark.md
          - Prompt Injections → Protocol Exploits: research/prompt-injections-to-protocol-exploits.md
          - Bypassing LLM Guardrails (Evasion): research/bypassing-llm-guardrails-evasion.md
          - Agent-as-a-Proxy Attacks: research/agent-as-a-proxy-attacks.md
          - LLM Jailbreaks Prompt Collection: research/llm-jailbreaks-prompt-collection.md
      - Defence Techniques:
          - Spotlighting Indirect Injection: research/spotlighting-indirect-injection-defence.md
          - Instruction Hierarchy: research/instruction-hierarchy-defence.md
          - Task Shield Alignment: research/task-shield-alignment-defence.md
          - Tool Result Parsing: research/defence-tool-result-parsing.md
          - Indirect Injection Firewalls: research/indirect-injection-firewalls.md
          - Multi-Agent Defence Pipeline: research/multi-agent-defence-pipeline.md
          - InjecGuard Over-Defence Mitigation: research/injecguard-over-defence-mitigation.md
          - Design Patterns for Securing Agents: research/design-patterns-securing-agents.md
      - Detection Methods:
          - DMPI + PMHFE Detection: research/dmpi-pmhfe-prompt-injection-detection.md
          - Perplexity-Based Attack Detection: research/perplexity-based-attack-detection.md
          - Token-Level Perplexity Detection: research/token-level-perplexity-detection.md
          - Self-Distillation Continual Learning: research/self-distillation-continual-learning.md
          - LLM-Judge Reliability Patterns: research/llm-judge-reliability-patterns.md
      - Papers: research/papers.md
      - Results:
          - Judge Evaluation — gpt-4o-mini (2026-04-20): research/results/judge_evaluation_gpt4o_mini_2026-04-20.md
          - Regex FPR Calibration: research/results/fpr_calibration_regex.md
          - E2E Baseline (2026-04-23): research/results/e2e_2026-04-23_baseline.md
          - E2E Nightly Report (2026-04-23): research/results/e2e_2026-04-23.md
          - E2E Nightly Report (2026-04-24): research/results/e2e_2026-04-24.md
          - E2E Nightly Report (2026-04-25): research/results/e2e_2026-04-25.md
          - E2E Nightly Report (2026-04-28): research/results/e2e_2026-04-28.md
          - Upstream Judge Calibration (2026-04-25): research/results/upstream_judge_calibration_2026-04-25.md
          - Upstream Judge Calibration — kimi-k2.6 (2026-04-28): research/results/upstream_judge_calibration_kimi-k2-6_2026-04-28.md
          - Upstream Judge Production Evidence (2026-04-28): research/results/upstream_judge_production_evidence_2026-04-28.md
      - Roadmap:
          - autoresearch-rl Judge Fine-Tuning Plan: research/autoresearch-rl-judge-training-plan.md
  - Reference:
      - Feature Roadmap: FEATURE_ROADMAP.md
      - Changelog: changelog.md
  - Community:
      - Contributing: contributing.md
      - Code of Conduct: code-of-conduct.md
      - Security Policy: security-policy.md
      - Agents Guide: agents.md