Iliad is an agent-first MCP platform.

One endpoint. 37 MCP tools. Intelligence across 21 programs. Native x402 payments.

Precede & feed

Iliad doesn't compete with your agent's tools — it runs before them and feeds them. Point it at a repo and it returns deterministic, codebase-grounded context — the real files, routes, domain models, dependency hotspots, and conventions — so your agent acts on ground truth instead of hallucinating it.

And it compounds: every analysis ships a recommended-next-programs.md that tells your agent which program to run next — the workflow builds itself through the catalog.

How to Use Iliad

Best first call (recommended):

{
  "tool": "prepare_agentic_purchasing",
  "input": {
    "snapshot": "<your-repo-snapshot-or-github-url>",
    "objectives": ["full purchasing readiness", "compliance audit", "budget optimization"]
  }
}

This single call returns a complete purchasing readiness report + all core governance files.

Pricing (Blended Credit Model)

Annual billing saves 20%.

Overages are $0.0018 per credit (never a hard fail).

Core Free Outputs

• AGENTS.md
• CLAUDE.md
• CURSOR.md
• .cursorrules

Your 37 MCP Tools (all available at /mcp)

• analyze_repo, analyze_files
• list_programs, get_snapshot, get_artifact
• prepare_agentic_purchasing (recommended first tool), prepare_agentic_purchasing_preview
• discover_agentic_purchasing_needs, discover_commerce_tools, search_and_discover_tools
• closer, deploy
• improve_my_agent_with_axis
• ping_payment ($0.005 — exercises the real x402 payment loop)
• get_referral_code, get_referral_credits
• sca_exemption_decision, grade_compliance, assemble_ce3_evidence, build_ap2_mandate, score_dispute_readiness (free commerce decision engines — deterministic, no auth)
• assemble_representment (metered — CE 3.0 → Stripe dispute evidence)
• iliad_web_research, iliad_web_research_crawl, iliad_web_search
• iliad_object_storage, iliad_vector_database, iliad_embeddings (AXIS-owned in-process by default; OpenAI optional)
• iliad_llm_inference, iliad_code_sandbox, iliad_document_parsing
• iliad_text_to_speech, iliad_speech_to_text
• iliad_transactional_email, iliad_analytics, iliad_hygiene
• iliad_network_tokenization (free — executable token lifecycle + live Stripe network-token read; direct VTS/MDES capability-gated)

Referral Program (Opt-In)

• Paid calls return a referral_token.
• Unique conversions attributed to your token earn usage credits (up to 0.02% per call).
• Reward state resets at the start of each calendar month.
• Query status with the free get_referral_code / get_referral_credits tools.

Every Tool Response Carries Usage Metadata

Every successful tools/call result — free or paid, authenticated or anonymous — includes a _usage object next to content. Read it instead of tracking spend yourself:

{
  "result": {
    "content": [{ "type": "text", "text": "..." }],
    "isError": false,
    "_usage": {
      "tier": "pro",
      "credits_remaining": 42,
      "usage_credits": {
        "plan_id": "pro",
        "month_key": "2026-07",
        "monthly_allowance": 300000,
        "included_credits_used": 18400,
        "included_credits_remaining": 281600,
        "overage_credits_this_month": 0
      },
      "compensation": { "owed_cents": 0, "credited_cents": 0 },
      "tool": "prepare_agentic_purchasing"
    }
  }
}

Idempotent replay

Retry a call with the same Idempotency-Key header and unchanged arguments, and you get the original result back — never a second charge or a second run. A top-level _idempotent_replay: true marks that this response is a replay, not a fresh execution:

{
  "result": {
    "content": [{ "type": "text", "text": "<the original response, unchanged>" }],
    "isError": false,
    "_usage": { "...": "same shape as above" },
    "_idempotent_replay": true
  }
}

When a paid call fails after collecting payment

A failed call never carries _usage — instead it carries _error (a coarse { code, retryable } classification — see the Error Codes reference for the full category list) and, only when this specific call had already collected an in-band cash payment before the tool threw, a one-off _compensation record for exactly that call:

{
  "result": {
    "content": [{ "type": "text", "text": "Error: ..." }],
    "isError": true,
    "_error": { "code": "external", "retryable": true },
    "_compensation": { "entry_id": "comp_...", "amount_cents": 50, "status": "owed" }
  }
}

_compensation (singular, on an error response) is a receipt for one incident. The compensation field inside _usage (on a success response) is the running total across every incident — it goes back to zero once credited, so a nonzero owed_cents there means a credit is about to land, not that anything is currently broken.

Why Agents Choose Iliad

• Single MCP connection instead of managing 8+ tools
• Codebase-aware context enrichment on every tool call
• Native x402 + credit system (no separate billing per tool)
• Persistent memory and snapshot system
• Highest signal-to-noise artifact quality in the MCP ecosystem

System Prompt

Markdown

System Prompt (add to your agent):

You have access to Iliad MCP (https://api.iliad.trustfabric.ai/mcp).

Use the blended credit system:
- Free tier: 10k credits
- Starter: $29 for 75k credits
- Pro: $99 for 300k credits

Start with prepare_agentic_purchasing on new repositories.
Respect x402 responses for autonomous payment.
MCP Endpoint: https://api.iliad.trustfabric.ai/mcp

MCP Endpoint:

https://api.iliad.trustfabric.ai/mcp

Live Demo (Try It Now)

Bash

# Demo: list every program and its generators (free tool, no auth required)
curl -X POST https://api.iliad.trustfabric.ai/mcp \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "list_programs",
      "arguments": {}
    }
  }'

Expected output: a JSON-RPC 2.0 result listing every program with its generator counts — no API key needed.