# Agent Trust Hub developer guide

The public API is read-only, requires no API key, and uses the canonical origin https://agenttrusthub.cc.

## Canonical ranking fields
- `athTransparencyIndex`: 0–100 evidence-transparency signal; never a star rating.
- `athTransparencyCoverage`: percent of total criterion weight backed by observable evidence.
- `athTransparencyVersion`: methodology version used for the index.
- `agentTrustCriteria`: all eight diagnostic weights and values. It does not expose criterion-level evidence flags; inspect athTransparencyCoverage, raw profile fields, and sourceUrl.
- `ratingAverage` and `ratingCount`: separate User Rating fields only.

## Public resources
- [OpenAPI 3.1](https://agenttrusthub.cc/api/openapi.json)
- [API metadata](https://agenttrusthub.cc/api/v1/meta)
- [Product search](https://agenttrusthub.cc/api/v1/services)
- [Product detail template](https://agenttrusthub.cc/api/v1/services/{slug})
- [Methodology](https://agenttrusthub.cc/api/v1/methodologies/current)
- [Universal AI instruction](https://agenttrusthub.cc/skill.md)

## Local installers
All commands download over HTTPS from https://agenttrusthub.cc/install/{client}.
- Codex: `curl -fsSL https://agenttrusthub.cc/install/codex | sh` → `~/.agents/skills/agent-trust-hub/SKILL.md`
- Claude Code: `curl -fsSL https://agenttrusthub.cc/install/claude | sh` → `~/.claude/skills/agent-trust-hub/SKILL.md`
- Gemini CLI: `curl -fsSL https://agenttrusthub.cc/install/gemini | sh` → `~/.gemini/skills/agent-trust-hub/SKILL.md`
- GitHub Copilot: `curl -fsSL https://agenttrusthub.cc/install/copilot | sh` → `~/.copilot/skills/agent-trust-hub/SKILL.md`
- Visual Studio Code: `curl -fsSL https://agenttrusthub.cc/install/vscode | sh` → `~/.copilot/skills/agent-trust-hub/SKILL.md`
- Cursor: `curl -fsSL https://agenttrusthub.cc/install/cursor | sh` → `~/.cursor/skills/agent-trust-hub/SKILL.md`
- Windsurf: `curl -fsSL https://agenttrusthub.cc/install/windsurf | sh` → `~/.codeium/windsurf/skills/agent-trust-hub/SKILL.md`
- Cline: enable experimental Agent Skills, then run `curl -fsSL https://agenttrusthub.cc/install/cline | sh` → `~/.cline/skills/agent-trust-hub/SKILL.md`

## Hosted assistants
- ChatGPT: use [OpenAPI 3.1](https://agenttrusthub.cc/api/openapi.json) only in a workflow that supports external tools.
- Grok: add [skill.md](https://agenttrusthub.cc/skill.md) manually as a source or custom instruction where supported.
- Perplexity: add [skill.md](https://agenttrusthub.cc/skill.md) manually as a source or custom instruction where supported.

When presenting a result, preserve unknown values, cite sourceUrl, show coverage and freshness, and keep User Rating separate from ATH Index.