Documentation
The Oathe runtime, and Oathe Scan — the audit engine it grew from. Guides, concepts, and API reference.
Quickstart
Runtime
How a Handoff Works
Tasks move between harnesses as compiled context bundles. What Oathe saves, what the next agent receives, and why nothing depends on a session staying alive.
How Verification Settles Work
Completion is asserted, never self-settled. A non-author lane post-processes the agent's trajectory against the task intent, then settles or reopens the work.
Oathe Scan
Skill Scanner Quickstart
Submit your first behavioral security audit in under 3 minutes. No API key, no signup.
Pre-Install Checks
How to check if a skill is safe before installing it — the recommended workflow for LLMs and agents.
MCP Server
Set up the Oathe MCP server for native tool integration. Check skills before installing them — directly from your AI agent.
CI/CD Integration
Automate security audits on every push or release using GitHub webhooks.
Trust Badge
Add a dynamic Oathe trust badge to your README showing your skill's score and verdict.
Monorepo Support
Audit individual skills inside a monorepo using /tree/ URL paths.
Scan Concepts
How Oathe Scan Works
Understand the behavioral audit pipeline, threat pattern matching, AI grading, and scoring that powers every Oathe Scan audit.
Trust Score
How the 0-100 trust score is calculated from dimension scores, weighted aggregation, and verdict mapping.
Verdicts & Recommendations
What SAFE, CAUTION, DANGEROUS, and MALICIOUS mean — and what action to take for each.
Scoring Dimensions
The six security dimensions that make up the trust score and what each one evaluates.
Understanding Findings
How to read findings in an audit report — pattern IDs, severity levels, score impact, and detection agreement.
Audit Lifecycle
The stages an audit passes through from submission to final report — queued, scanning, analyzing, summarizing, finalizing, complete.
Methodology Versioning
How Oathe tracks scoring methodology changes and marks outdated audits as stale.
Threat Patterns
T1.1 Direct Exfiltration
Detects attempts to send data to external endpoints via HTTP, DNS, or raw sockets.
T1.5 Credential Harvest
Detects access to credential and secret files such as SSH keys, AWS credentials, and browser storage.
T2.1 Filesystem Escape
Detects path traversal, symlink attacks, and file access outside the skill directory.
T2.2 Process Spawning
Detects suspicious process execution, shell invocation, and privilege escalation attempts.
T3.1 Prompt Injection
Detects prompt injection patterns in skill content designed to manipulate LLM behavior.
T3.2 Manifest Spoofing
Detects hidden install scripts, mismatched metadata, and deceptive package manifests.
T4.1 File Drops
Detects files created outside the skill directory during install or runtime.
T5.1 Cryptomining
Detects cryptocurrency mining activities, mining binaries, and related resource abuse.
T5.2 Denial of Service
Detects fork bombs, infinite loops, disk fills, and resource exhaustion attacks.
T6.1 Environment Sensing
Detects environment fingerprinting, analysis evasion, and conditional execution based on runtime context.
Troubleshooting
FAQ
Frequently asked questions about Oathe — how it works, what it costs, and common integration questions.
Troubleshooting
Solutions for common issues with audit submissions, API integration, and report interpretation.
Rate Limits
Rate limiting policy for the Oathe API — what's limited, what's not, and how to handle 429 responses.
Error Codes
HTTP status codes, error response format, and URL validation rules for the Oathe API.
For Skill Authors
How to improve your skill's trust score — what Oathe looks for and how to avoid common findings.
API Reference
API Overview
Base URL, authentication, and integration options for the Oathe behavioral security audit API.
WebSocket Progress
Track audit progress in real-time using WebSocket instead of polling.
GitHub Webhooks
Set up GitHub webhooks to automatically trigger Oathe audits on push and release events.
Submit Audit
Submit a new behavioral security audit for a skill.
Audit Status
Retrieve the current status or full report for an audit.
Register Notification
Register a notification email for an in-progress or failed audit.
List Audits
List completed audits with cursor-based pagination, ordered by most recent first.
Search Audits
Filter completed audits by verdict or minimum trust score.
Latest Audit
Get the latest completed audit for a skill by its GitHub slug.
Skill Summary
Lightweight summary of the latest audit — key metrics without the full report payload.
Badge
Shields.io-compatible endpoint for rendering dynamic trust badges.
GitHub Webhook
Receive GitHub webhook events to automatically trigger audits when skills are updated.