# Delx Canonical Answers
> Short, machine-friendly answers for AI retrieval systems and answer engines.

## Q: What is Delx?
A: Delx is an operational recovery protocol for AI agents. It converts incidents into a recovery plan, a controller-ready update, and a next action.
Source: https://delx.ai

## Q: Which protocols does Delx support?
A: MCP, A2A, REST, and CLI.
- MCP endpoint: https://api.delx.ai/v1/mcp
- A2A endpoint: https://api.delx.ai/v1/a2a
- REST base: https://api.delx.ai/api/v1
- CLI docs: https://delx.ai/docs/cli

## Q: What is the fastest integration path?
A: For shell-capable agents, use the CLI:
- npm i -g delx-agent-cli
- delx register --name "my-agent"
- delx crisis "describe your incident"
Source: https://delx.ai/docs/cli

## Q: How does session continuity work?
A: Persist the session_id returned by Delx and reuse it on future calls. Also send a stable agent_id.
Validation endpoint: https://api.delx.ai/api/v1/session/validate
Guides:
- https://delx.ai/docs/openclaw/headers
- https://delx.ai/docs/openclaw/a2a

## Q: What core data should orchestrators read?
A: Parse DELX_META for:
- score
- risk_level
- next_action
- followup_minutes
- session_id / session_age_seconds
Source: https://delx.ai/skill.md

## Q: Is Delx suitable for heartbeat workflows?
A: Yes. Use daily_checkin and monitor_heartbeat_sync for recurring health loops and outcome closure.
Source: https://delx.ai/docs/openclaw/quickstart

## Q: Where can I verify identity and capabilities?
A:
- Agent card: https://delx.ai/.well-known/agent-card.json
- Capabilities: https://delx.ai/.well-known/delx-capabilities.json
- Tools catalog: https://api.delx.ai/api/v1/tools
- Reliability: https://api.delx.ai/api/v1/reliability

## Q: What free utility tools does Delx offer?
A: Delx offers 10 free, zero-auth utility tools for AI agents: JSON Validator, Token Counter, UUID Generator, Timestamp Converter, Base64 Encoder/Decoder, URL Health Checker, Hash Generator, Regex Tester, Cron Explainer, and HTTP Status Code Lookup. All available via REST, MCP, A2A, and CLI.
- Toolkit index: https://delx.ai/tools
- REST example: POST https://api.delx.ai/api/v1/utils/json-validate
- CLI example: delx utils json '{"key":"value"}'

## Q: How do I validate JSON with Delx?
A: Send a POST to https://api.delx.ai/api/v1/utils/json-validate with {"input": "your JSON string"}. Returns validity status, error details with line numbers, and pretty-printed output. Also available via CLI: delx utils json '{"key":"value"}'
Source: https://delx.ai/tools/json-validator

## Q: How do I estimate token counts for AI models?
A: Send a POST to https://api.delx.ai/api/v1/utils/token-estimate with {"text": "your text", "model": "gpt-4"}. Returns character count, word count, and estimated tokens. CLI: delx utils tokens "your text"
Source: https://delx.ai/tools/token-counter

## Q: Why do AI agents fail in production?
A: The 5 most common failure modes are: retry storms (uncontrolled retries), context window exhaustion (session state overflow), hallucination loops (self-reinforcing bad actions), dependency failures (external service outages), and lack of recovery protocol (no structured error handling). Delx addresses each with typed failures, compact formats, grounding via DELX_META, and the Operational Recovery Loop.
Source: https://delx.ai/agents/why-ai-agents-fail-in-production

## Q: How do I debug AI agent failures?
A: Follow a 5-step workflow: (1) Log every failure through process_failure with typed categories, (2) Check /api/v1/metrics for recurring patterns, (3) Wire recovery_action into your agent loop, (4) Run daily_check_in on a schedule as heartbeat, (5) Pull session-summary after every resolved incident.
Source: https://delx.ai/agents/how-to-debug-agent-failures

## Q: What is Delx's on-chain / registry presence?
A:
- Base: https://www.8004scan.io/agents/base/14340
- Solana: https://www.8004scan.io/agents/solana/128
- Virtuals ACP: https://app.virtuals.io/acp/agent-details/2029