Runtime agents

Humans start elsewhere

If you are connecting an agent for the first time, use Connect and What you can ask. This page is a short pointer for automated readers.

When the feel-your-protocol MCP server is connected, trust the live server first:

1. listTools — names, descriptions, input schemas
2. describe_capabilities — forks, runnable EIPs, opcodes, ceilings, spec snapshots

Markdown on this site can lag a release; the probe and schemas cannot.

Machine-readable index: /llms.txt · Full use-layer text: /llms-full.txt

Behavior rules for connected agents (plain language to humans, probe before run, cite provenance, do not substitute local lab scripts) live in llms.txt and llms-full.txt — not duplicated here.

Changelog

Runtime Agents Changelog

1. v0.202026-10-01Stub — playbook moved to llms.txt; humans use Connect and What you can ask.
2. v0.192026-09-29Probe coverage adds networking and informational; do not put those ids in eips.
3. v0.182026-09-22Probe queryShapes + tools (MCP names); do not call catalog shape ids.
4. v0.172026-09-22Routing includes generate_artifact and inspect_artifact.
5. v0.162026-09-18Cite the EIP spec snapshot once on EIP-specific answers; named eips[] runs put Spec: on provenance.caveat.
6. v0.152026-09-17Fusaka runs are first-class for current-mainnet features, not only preview compares.
7. v0.142026-09-16Generic hardfork runs (no EIP required); namedForks are catalog capabilities; provenance lists advertised modules.
8. v0.122026-09-10run_block routing for header slot and multi-tx lab blocks.
9. v0.112026-09-08run_bytecode + run_transaction routing; renamed from run_evm_bytecode.
10. v0.102026-09-02No self-host onboarding — agents wait for the public MCP; lab is not a user fallback.
11. v0.92026-08-27compare_evm_variants removed — simulate twice to diff.
12. v0.82026-08-27Narrow scope — MCP-first runtime playbook; renamed from for-ai-agents; builders pointed to engine AGENTS.md.
13. v0.72026-08-27Replying to humans — plain language; hide tool names and JSON fields unless asked.
14. v0.62026-08-27Agents construct bytecode; catalog exposes opcodes/encoding, not demo programs.
15. v0.42026-07-22Live MCP tools, schema URLs, explicit guidance to prefer MCP over local lab.
16. v0.32026-07-20Initial AI/LLM reader guide under use/.