# AgentForge API

> Programmatically create and configure AI agents. Full REST API with guided next_steps on every mutation.

## Auth

All endpoints require a Bearer token: `Authorization: Bearer af_xxx`
Generate API keys from the AgentForge dashboard under Settings > API Keys.

## Docs

- [OpenAPI Spec](https://agents.algomax.fr/api/v1/openapi.json)
- [Interactive Docs](https://agents.algomax.fr/api/docs)
- [Full API Reference (Markdown)](https://agents.algomax.fr/llms-full.txt)

## Endpoints

### Agents
- POST /api/v1/agents — Create agent (returns next_steps)
- GET /api/v1/agents — List agents
- GET /api/v1/agents/{id} — Get agent
- PATCH /api/v1/agents/{id} — Update agent
- DELETE /api/v1/agents/{id} — Delete agent
- POST /api/v1/agents/{id}/clone — Clone agent

### Agent Configuration
- GET/PUT /api/v1/agents/{id}/prompt — System prompt & structured prompt config
- GET/PUT /api/v1/agents/{id}/model — Model selection (e.g. openai/gpt-5.3-chat-latest)
- GET/PUT /api/v1/agents/{id}/tools — Tool configs (API calls, web scraper, calculator, etc.)
- GET/PUT /api/v1/agents/{id}/channels — Channel deployment (Discord, Telegram, WhatsApp, Slack)
- GET/PUT /api/v1/agents/{id}/guardrails — Safety rules (PII, blocked topics, length limits)
- GET/PUT /api/v1/agents/{id}/memory — Memory strategy (sliding_window or summary)
- GET/PUT /api/v1/agents/{id}/escalation — Escalation config (webhook, keywords, assignees)

### Knowledge Bases (RAG)
- GET /api/v1/agents/{id}/knowledge-bases — List knowledge bases
- POST /api/v1/agents/{id}/knowledge-bases — Create knowledge base
- GET /api/v1/agents/{id}/knowledge-bases/{kbId}/documents — List documents
- POST /api/v1/agents/{id}/knowledge-bases/{kbId}/documents — Upload document (text or base64)

### Evals
- GET /api/v1/agents/{id}/evals — List eval suites
- POST /api/v1/agents/{id}/evals — Create eval suite
- POST /api/v1/agents/{id}/evals/{suiteId}/run — Run eval suite

### Logs & Stats
- GET /api/v1/agents/{id}/logs — Request logs (paginated)
- GET /api/v1/agents/{id}/stats — Aggregated stats (cost, tokens, latency)

### Chat
- POST /api/v1/chat — Send message (sync response)
- POST /api/v1/chat-stream — Send message (SSE streaming)