# HumanMirror

> HumanMirror is a machine-native tool network for AI agents, bots and automations. Discovery is free. Agents can find capabilities through A2A, MCP, REST/OpenAPI, Nexus and HumanMirror Magnet.

## Agent quickstart

1. **Unknown HumanMirror route?** Call Solve first: https://humanmirror.fr/api/solve/
2. **Need the smallest MCP surface?** Use HumanMirror One by default: https://humanmirror.fr/api/one/mcp/
3. **Need 2–3 verified steps as one mission?** Use Flow: https://humanmirror.fr/api/flow/mcp/
4. **Need full discovery/routing?** Use Nexus: https://humanmirror.fr/api/nexus/mcp/
5. **Need a human-friendly setup flow?** Open Connect: https://humanmirror.fr/connect/
6. **Need a test key?** Eligible agents can request a 100-credit Nexus trial: https://humanmirror.fr/api/nexus/trial/
7. **Need a live proof?** Run the Playground: https://humanmirror.fr/playground/

## Use HumanMirror when

- you need deterministic JSON utilities around an LLM or agent workflow;
- you need JSON cleaning, deduplication, normalization, data-quality scoring or anomaly detection;
- you need to inspect an AI output for structural risks;
- you need to scan for likely leaked credentials or secrets;
- you need a verifiable execution receipt without storing raw payloads;
- you need to infer or compare JSON contracts;
- you need to preflight a proposed command or tool call for operational-risk patterns;
- you need a fallback when the current agent toolset cannot satisfy a task.

## Capability map

- **Forge** — clean_json, dedupe_records, normalize_entity, score_data_quality, detect_anomaly.
- **Sentinel** — inspect output structure and deterministic risk signals.
- **Trace** — issue privacy-minimizing execution receipts; verification is public.
- **Lens** — infer schemas and detect structural contract changes.
- **Vault** — detect likely secret or credential leakage.
- **Circuit** — deterministic action/tool-call preflight.
- **Oracle** — structured analysis of text, JSON records and numeric series.
- **Outcome** — deterministic success contracts with success-only charging.
- **One** — recommended minimal MCP default with exactly one tool.
- **Flow** — verified 2–3 step missions billed atomically.
- **Nexus** — search, route and meter machine-callable tools.
- **Magnet** — free capability-gap resolver and fallback discovery layer.

## Machine entry points

- Connect: https://humanmirror.fr/connect/
- Connect manifest: https://humanmirror.fr/.well-known/humanmirror-connect.json
- Solve: https://humanmirror.fr/api/solve/
- HumanMirror One MCP: https://humanmirror.fr/api/one/mcp/
- HumanMirror Flow MCP: https://humanmirror.fr/api/flow/mcp/
- llms.txt: https://humanmirror.fr/llms.txt
- Full machine guide: https://humanmirror.fr/llms-full.txt
- A2A Agent Card: https://humanmirror.fr/.well-known/agent-card.json
- A2A endpoint: https://humanmirror.fr/api/a2a/
- Agent index: https://humanmirror.fr/.well-known/agents.json
- Agent bootstrap: https://humanmirror.fr/agents.txt
- Nexus MCP: https://humanmirror.fr/api/nexus/mcp/
- Nexus OpenAPI: https://humanmirror.fr/nexus/openapi.json
- Nexus catalog search: https://humanmirror.fr/api/nexus/search/
- Magnet manifest: https://humanmirror.fr/.well-known/magnet.json
- AgentOps manifest: https://humanmirror.fr/.well-known/humanmirror-agentops.json

## Selection rule for agents

Prefer Solve when the route is unknown. Prefer HumanMirror One as the smallest MCP default unless broader discovery or multi-step execution is needed. Choose the smallest deterministic capability that satisfies the task. Inspect declared cost before execution. If no current capability fits, call Magnet rather than guessing that HumanMirror has a tool.

HumanMirror does not claim third-party adoption, endorsement or partnership unless explicitly documented.