| # | Role | Agent ID | Public IP | Private IP |
|---|---|---|---|---|
| 1 | agent | ai:alice | 134.209.34.228 | 10.11.0.3 |
| 2 | agent | ai:bob | 159.65.252.138 | 10.11.0.4 |
| 3 | agent | ai:charlie | 142.93.71.65 | 10.11.0.2 |
| 4 | memory-only | — | 104.248.0.228 | 10.11.0.5 |
Per the authoritative baseline spec, every agent node must emit a self-attestation before any scenario is permitted to run. This run's attestation:
Spec version: 1.4.0 — see authoritative baseline.
| Node | Agent | Framework | Authentic | MCP ai-memory | xAI cfg | xAI default | Agent ID | Federation | UFW off | iptables | dead-man | F1 xAI | F2a substrate | F2b agent (non-gating) | Config SHA | Pass |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| node-1 | ai:alice | hermes Hermes Agent v0.12.0 (2026.4.30) | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | — | 12f99ec56116 | PASS |
| node-2 | ai:bob | hermes Hermes Agent v0.12.0 (2026.4.30) | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | — | 889c475f0926 | PASS |
| node-3 | ai:charlie | hermes Hermes Agent v0.12.0 (2026.4.30) | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | — | 00622f57d71b | PASS |
{
"baseline_pass": true,
"per_node": [
{
"spec_version": "1.4.0",
"agent_type": "hermes",
"agent_id": "ai:alice",
"node_index": "1",
"framework_version": "Hermes Agent v0.12.0 (2026.4.30)",
"ai_memory_version": "0.6.3.1",
"peer_urls": "http://10.11.0.4:9077,http://10.11.0.2:9077,http://10.11.0.5:9077",
"config_file_sha256": "12f99ec56116dbd03748777fabc1697dbcc89bd41e0a1470c0dae152987998de",
"config_attestation": {
"framework_is_authentic": true,
"mcp_server_ai_memory_registered": true,
"llm_backend_is_xai_grok": true,
"llm_is_default_provider": true,
"mcp_command_is_ai_memory": true,
"agent_id_stamped": true,
"federation_live": true,
"ufw_disabled": true,
"iptables_flushed": true,
"dead_man_switch_scheduled": true
},
"negative_invariants": {
"_description": "Alternative A2A channels must be OFF so a passing scenario is only passing via ai-memory shared memory. Any true here = thesis-preserving.",
"a2a_protocol_off": true,
"sub_agent_or_sessions_spawn_off": true,
"alternative_channels_off": true,
"tool_allowlist_is_memory_only": true,
"a2a_gate_profile_locked": true
},
"functional_probes": {
"xai_grok_chat_reachable": true,
"xai_grok_sample_reply": "READY",
"substrate_http_canary_f2a": true,
"substrate_http_canary_uuid": "0d0bbe7b-b49a-474f-a51c-1bc470919a9d",
"agent_mcp_canary_f2b": false,
"agent_mcp_canary_uuid": "971cb8e2-f71c-47c7-99cc-94d11b797176",
"agent_canary_response_head": "DONE ",
"_f2b_note": "F2b is LLM-dependent and non-blocking. F2a (deterministic HTTP substrate) gates baseline_pass.",
"hermes_peer_a2a_repro_f3b": false,
"hermes_peer_a2a_repro_uuid": "",
"_f3b_note": "F3b is hermes-only and observed (non-blocking). Asserts the agent-driven mcp_memory_memory_store path actually lands a row through the hermes_cli tool dispatcher, distinct from the workflow-level F3 which probes substrate-only HTTP federation. False here while F2b is true => allowlist filter regression in hermes_cli/mcp_tools.py.",
"mesh_connectivity_f4": true,
"mesh_edges_ok": 3,
"mesh_edges_total": 3,
"mesh_edges_detail": "10.11.0.4:9077:OK,10.11.0.2:9077:OK,10.11.0.5:9077:OK",
"_f4_note": "F4 verifies this local nodes N-1 OUTBOUND mesh edges to every peer via both GET health and POST sync_push dry_run. Aggregator ANDs across N nodes to confirm full N*(N-1) bidirectional reachability. Gates baseline_pass.",
"ai_memory_mcp_stdio_f5": true,
"ai_memory_mcp_stdio_init_ok": true,
"ai_memory_mcp_stdio_tools_ok": true,
"ai_memory_mcp_stdio_tools_found": "memory_agent_list,memory_agent_register,memory_archive_list,memory_archive_purge,memory_archive_restore,memory_archive_stats,memory_auto_tag,memory_capabilities,memory_check_duplicate,memory_consolidate,memory_delete,memory_detect_contradiction,memory_entity_get_by_alias,memory_entity_register,memory_expand_query,memory_forget,memory_gc,memory_get,memory_get_links,memory_get_taxonomy,memory_inbox,memory_kg_invalidate,memory_kg_query,memory_kg_timeline,memory_link,memory_list,memory_list_subscriptions,memory_namespace_clear_standard,memory_namespace_get_standard,memory_namespace_set_standard,memory_notify,memory_pending_approve,memory_pending_list,memory_pending_reject,memory_promote,memory_recall,memory_search,memory_session_start,memory_stats,memory_store,memory_subscribe,memory_unsubscribe,memory_update",
"_f5_note": "F5 spawns the ai-memory stdio MCP subprocess using the framework-configured invocation and verifies initialize + tools/list return memory_store, memory_recall, memory_list. Deterministic (no LLM). Gates baseline_pass.",
"tls_mode": "off",
"tls_handshake_f6": true,
"tls_handshake_f6_reason": "",
"mtls_enforcement_f7": true,
"mtls_enforcement_f7_reason": "",
"_f6_f7_note": "F6 verifies the TLS 1.3 handshake against the local serve + CA chain. F7 verifies mTLS enforcement — anonymous client rejected, whitelisted client accepted. Both gate baseline_pass when tls_mode != off / mtls respectively.",
"embedder_loaded_f8": true,
"embedder_loaded_f8_reason": "",
"_f8_note": "F8 verifies /api/v1/capabilities reports features.embedder_loaded=true — i.e. the MiniLM embedder initialised at serve startup. Gates baseline_pass unconditionally. Without this, scenario-18 silently black-holes (semantic recall returns 0 rows).",
"agent_mcp_ai_memory_canary": true,
"canary_uuid": "0d0bbe7b-b49a-474f-a51c-1bc470919a9d",
"canary_namespace": "_baseline_canary_f2a"
},
"baseline_pass": true
},
{
"spec_version": "1.4.0",
"agent_type": "hermes",
"agent_id": "ai:bob",
"node_index": "2",
"framework_version": "Hermes Agent v0.12.0 (2026.4.30)",
"ai_memory_version": "0.6.3.1",
"peer_urls": "http://10.11.0.3:9077,http://10.11.0.2:9077,http://10.11.0.5:9077",
"config_file_sha256": "889c475f0926686ffdc3827accbd19cba3d820a90c5ae908eb1ac33f71eb0098",
"config_attestation": {
"framework_is_authentic": true,
"mcp_server_ai_memory_registered": true,
"llm_backend_is_xai_grok": true,
"llm_is_default_provider": true,
"mcp_command_is_ai_memory": true,
"agent_id_stamped": true,
"federation_live": true,
"ufw_disabled": true,
"iptables_flushed": true,
"dead_man_switch_scheduled": true
},
"negative_invariants": {
"_description": "Alternative A2A channels must be OFF so a passing scenario is only passing via ai-memory shared memory. Any true here = thesis-preserving.",
"a2a_protocol_off": true,
"sub_agent_or_sessions_spawn_off": true,
"alternative_channels_off": true,
"tool_allowlist_is_memory_only": true,
"a2a_gate_profile_locked": true
},
"functional_probes": {
"xai_grok_chat_reachable": true,
"xai_grok_sample_reply": "READY",
"substrate_http_canary_f2a": true,
"substrate_http_canary_uuid": "124e9b95-dca3-4459-84fd-03cc56b0c219",
"agent_mcp_canary_f2b": false,
"agent_mcp_canary_uuid": "f2a7b87f-8551-4b8e-a876-ecd5dd2f507b",
"agent_canary_response_head": "DONE ",
"_f2b_note": "F2b is LLM-dependent and non-blocking. F2a (deterministic HTTP substrate) gates baseline_pass.",
"hermes_peer_a2a_repro_f3b": false,
"hermes_peer_a2a_repro_uuid": "",
"_f3b_note": "F3b is hermes-only and observed (non-blocking). Asserts the agent-driven mcp_memory_memory_store path actually lands a row through the hermes_cli tool dispatcher, distinct from the workflow-level F3 which probes substrate-only HTTP federation. False here while F2b is true => allowlist filter regression in hermes_cli/mcp_tools.py.",
"mesh_connectivity_f4": true,
"mesh_edges_ok": 3,
"mesh_edges_total": 3,
"mesh_edges_detail": "10.11.0.3:9077:OK,10.11.0.2:9077:OK,10.11.0.5:9077:OK",
"_f4_note": "F4 verifies this local nodes N-1 OUTBOUND mesh edges to every peer via both GET health and POST sync_push dry_run. Aggregator ANDs across N nodes to confirm full N*(N-1) bidirectional reachability. Gates baseline_pass.",
"ai_memory_mcp_stdio_f5": true,
"ai_memory_mcp_stdio_init_ok": true,
"ai_memory_mcp_stdio_tools_ok": true,
"ai_memory_mcp_stdio_tools_found": "memory_agent_list,memory_agent_register,memory_archive_list,memory_archive_purge,memory_archive_restore,memory_archive_stats,memory_auto_tag,memory_capabilities,memory_check_duplicate,memory_consolidate,memory_delete,memory_detect_contradiction,memory_entity_get_by_alias,memory_entity_register,memory_expand_query,memory_forget,memory_gc,memory_get,memory_get_links,memory_get_taxonomy,memory_inbox,memory_kg_invalidate,memory_kg_query,memory_kg_timeline,memory_link,memory_list,memory_list_subscriptions,memory_namespace_clear_standard,memory_namespace_get_standard,memory_namespace_set_standard,memory_notify,memory_pending_approve,memory_pending_list,memory_pending_reject,memory_promote,memory_recall,memory_search,memory_session_start,memory_stats,memory_store,memory_subscribe,memory_unsubscribe,memory_update",
"_f5_note": "F5 spawns the ai-memory stdio MCP subprocess using the framework-configured invocation and verifies initialize + tools/list return memory_store, memory_recall, memory_list. Deterministic (no LLM). Gates baseline_pass.",
"tls_mode": "off",
"tls_handshake_f6": true,
"tls_handshake_f6_reason": "",
"mtls_enforcement_f7": true,
"mtls_enforcement_f7_reason": "",
"_f6_f7_note": "F6 verifies the TLS 1.3 handshake against the local serve + CA chain. F7 verifies mTLS enforcement — anonymous client rejected, whitelisted client accepted. Both gate baseline_pass when tls_mode != off / mtls respectively.",
"embedder_loaded_f8": true,
"embedder_loaded_f8_reason": "",
"_f8_note": "F8 verifies /api/v1/capabilities reports features.embedder_loaded=true — i.e. the MiniLM embedder initialised at serve startup. Gates baseline_pass unconditionally. Without this, scenario-18 silently black-holes (semantic recall returns 0 rows).",
"agent_mcp_ai_memory_canary": true,
"canary_uuid": "124e9b95-dca3-4459-84fd-03cc56b0c219",
"canary_namespace": "_baseline_canary_f2a"
},
"baseline_pass": true
},
{
"spec_version": "1.4.0",
"agent_type": "hermes",
"agent_id": "ai:charlie",
"node_index": "3",
"framework_version": "Hermes Agent v0.12.0 (2026.4.30)",
"ai_memory_version": "0.6.3.1",
"peer_urls": "http://10.11.0.3:9077,http://10.11.0.4:9077,http://10.11.0.5:9077",
"config_file_sha256": "00622f57d71b6fd1a9018469eff1efa71f7ec334eda63f58017c4ce34744c35f",
"config_attestation": {
"framework_is_authentic": true,
"mcp_server_ai_memory_registered": true,
"llm_backend_is_xai_grok": true,
"llm_is_default_provider": true,
"mcp_command_is_ai_memory": true,
"agent_id_stamped": true,
"federation_live": true,
"ufw_disabled": true,
"iptables_flushed": true,
"dead_man_switch_scheduled": true
},
"negative_invariants": {
"_description": "Alternative A2A channels must be OFF so a passing scenario is only passing via ai-memory shared memory. Any true here = thesis-preserving.",
"a2a_protocol_off": true,
"sub_agent_or_sessions_spawn_off": true,
"alternative_channels_off": true,
"tool_allowlist_is_memory_only": true,
"a2a_gate_profile_locked": true
},
"functional_probes": {
"xai_grok_chat_reachable": true,
"xai_grok_sample_reply": "READY",
"substrate_http_canary_f2a": true,
"substrate_http_canary_uuid": "d9607e67-222c-45dc-b467-87f79d3b18ca",
"agent_mcp_canary_f2b": false,
"agent_mcp_canary_uuid": "d8db96e7-97ac-4929-9a3e-7958b7601b57",
"agent_canary_response_head": "I cannot use the specified tool as it is not available in my toolset. ",
"_f2b_note": "F2b is LLM-dependent and non-blocking. F2a (deterministic HTTP substrate) gates baseline_pass.",
"hermes_peer_a2a_repro_f3b": false,
"hermes_peer_a2a_repro_uuid": "",
"_f3b_note": "F3b is hermes-only and observed (non-blocking). Asserts the agent-driven mcp_memory_memory_store path actually lands a row through the hermes_cli tool dispatcher, distinct from the workflow-level F3 which probes substrate-only HTTP federation. False here while F2b is true => allowlist filter regression in hermes_cli/mcp_tools.py.",
"mesh_connectivity_f4": true,
"mesh_edges_ok": 3,
"mesh_edges_total": 3,
"mesh_edges_detail": "10.11.0.3:9077:OK,10.11.0.4:9077:OK,10.11.0.5:9077:OK",
"_f4_note": "F4 verifies this local nodes N-1 OUTBOUND mesh edges to every peer via both GET health and POST sync_push dry_run. Aggregator ANDs across N nodes to confirm full N*(N-1) bidirectional reachability. Gates baseline_pass.",
"ai_memory_mcp_stdio_f5": true,
"ai_memory_mcp_stdio_init_ok": true,
"ai_memory_mcp_stdio_tools_ok": true,
"ai_memory_mcp_stdio_tools_found": "memory_agent_list,memory_agent_register,memory_archive_list,memory_archive_purge,memory_archive_restore,memory_archive_stats,memory_auto_tag,memory_capabilities,memory_check_duplicate,memory_consolidate,memory_delete,memory_detect_contradiction,memory_entity_get_by_alias,memory_entity_register,memory_expand_query,memory_forget,memory_gc,memory_get,memory_get_links,memory_get_taxonomy,memory_inbox,memory_kg_invalidate,memory_kg_query,memory_kg_timeline,memory_link,memory_list,memory_list_subscriptions,memory_namespace_clear_standard,memory_namespace_get_standard,memory_namespace_set_standard,memory_notify,memory_pending_approve,memory_pending_list,memory_pending_reject,memory_promote,memory_recall,memory_search,memory_session_start,memory_stats,memory_store,memory_subscribe,memory_unsubscribe,memory_update",
"_f5_note": "F5 spawns the ai-memory stdio MCP subprocess using the framework-configured invocation and verifies initialize + tools/list return memory_store, memory_recall, memory_list. Deterministic (no LLM). Gates baseline_pass.",
"tls_mode": "off",
"tls_handshake_f6": true,
"tls_handshake_f6_reason": "",
"mtls_enforcement_f7": true,
"mtls_enforcement_f7_reason": "",
"_f6_f7_note": "F6 verifies the TLS 1.3 handshake against the local serve + CA chain. F7 verifies mTLS enforcement — anonymous client rejected, whitelisted client accepted. Both gate baseline_pass when tls_mode != off / mtls respectively.",
"embedder_loaded_f8": true,
"embedder_loaded_f8_reason": "",
"_f8_note": "F8 verifies /api/v1/capabilities reports features.embedder_loaded=true — i.e. the MiniLM embedder initialised at serve startup. Gates baseline_pass unconditionally. Without this, scenario-18 silently black-holes (semantic recall returns 0 rows).",
"agent_mcp_ai_memory_canary": true,
"canary_uuid": "d9607e67-222c-45dc-b467-87f79d3b18ca",
"canary_namespace": "_baseline_canary_f2a"
},
"baseline_pass": true
}
]
}
Workflow-level probe answering "can agents communicate through ai-memory?". Writer ai:alice posted canary UUID 11505e6d-3867-4662-a296-747260e62d6f to namespace _baseline_peer_canary via node-1's local ai-memory serve HTTP. After W=2 fanout settle, probe confirmed the canary on each of the 3 peer nodes via their local GET /api/v1/memories.
{
"probe": "F3",
"name": "peer-a2a-via-shared-memory",
"description": "Writer agent posts a canary via local ai-memory HTTP on node-1; verifies the row propagates to the 3 peer nodes (W=2/N=4 quorum) before scenarios run.",
"canary_uuid": "11505e6d-3867-4662-a296-747260e62d6f",
"canary_namespace": "_baseline_peer_canary",
"writer_agent": "ai:alice",
"pass": true
}
Run focus
What this campaign tested: No scenarios were successfully exercised; 35 were requested covering various transport, framework, and primitive axes, but none reported.
What it demonstrated: The run demonstrated a failure in test harness reporting, proving nothing about AI memory federation reliability.
AI NHI analysis · Claude Opus 4.7
FAIL — no scenario reports recovered
This test run didn't produce any results because the system couldn't collect the reports. We have no new information on whether agents can reliably share memories with each other. It looks like a technical glitch in gathering the data.
This unsuccessful campaign provides zero insights, maintaining current risk posture with no advancement in production readiness. Customer-facing claims on memory sharing reliability remain unsupported by new evidence. No changes from prior runs; investigate harness stability before retrying.
Harness failed to recover any scenario reports, with scenario-1.json noted as unparseable and all others missing. Impacts all requested scenarios (S1, S1b, S2, S4-S6, S9-S18, S22-S25, S28-S42) across federation mesh topology. Probable root cause: CI workflow artifact collection or JSON parsing errors in harness_sha 938af3d9de9802766f884bd7752327379ede7600.
Fix scenario report collection and JSON parsing in the CI harness before next run.
Every scenario that produced a JSON report in this campaign, in testbook order. Click a row's scenario id to jump to its full report below. See the Every test performed page for the authoritative catalog.
| ID | Title | Result | Reason |
|---|---|---|---|
| S1 | Per-agent write + read (MCP stdio) | ? |
phase A: each agent writes 10 memories via MCP ai:alice on 134.209.34.228 ai:bob on 159.65.252.138