v0.7.0 A2A · 2-round consecutive 100% GREEN gate · openclaw ↔ hermes via Grok 4.2

v0.7.0 A2A campaign — v0.7.0-a2a-wave4-r2-20260509-1545

Two 16 GB DigitalOcean droplets · ai-memory v0.7.0 · grok-4.20-0309-reasoning

Round 1BLOCK
Round 2BLOCK

Generated 2026-05-09 16:04 UTC.

Scenarios planned
79
In-scope
74
Passed
44
Failed
22
Skipped
13
Agents
2
Droplets
2
Audit chain
Per-scenario evidence

All scenarios in v0.7.0-a2a-wave4-r2-20260509-1545

S1SKIP

 

S10PASS

 

S11PASS

 

S12PASS

 

S13PASS

 

S14SKIP

 

S15PASS

 

S16FAIL

promote endpoint returned HTTP 404; bob sees tier="short", expected "long"

S17PASS

 

S18FAIL

semantic query did not surface alice's memory; semantic query did not surface bob's memory

S1bPASS

 

S2PASS

 

S20SKIP

scenario 20 only runs under tls_mode=mtls (actual: off)

S21SKIP

scenario 21 only runs under tls_mode=mtls (actual: off)

S22PASS

 

S23PASS

 

S24PASS

 

S25SKIP

 

S26SKIP

scenario 26 only runs under agent_group=mixed (actual: openclaw_hermes)

S27SKIP

 

S28PASS

 

S29PASS

 

S30PASS

 

S31PASS

 

S32FAIL

bob's inbox did not deliver alice's notify

S33FAIL

bob's subscription list did not include the subscribed namespace

S34FAIL

charlie did not see approved row

S35FAIL

get-standard returned HTTP 501; parent rule not layered into child's standard view; child rule missing from standard view

S36PASS

 

S37PASS

 

S38PASS

 

S39SKIP

 

S4PASS

 

S40PASS

 

S41PASS

 

S42PASS

 

S43PASS

 

S44FAIL

could not locate scenario44-955e0c/alphaone in taxonomy response; root node missing subtree_count field — v0.6.3 should expose it

S45FAIL

kg_query(past) returned HTTP 503; kg_query(now) returned HTTP 503; kg_invalidate returned HTTP 503; as_of=past missing edges; expected all 3 targets, got 0; as_of=now expected 2 edges, got 0

S46FAIL

timeline endpoint returned HTTP 503; expected >=2 edge events (M0->T1 invalidated + M0->T2 live), got 0; no live edge in timeline; no invalidated edge in timeline (valid_until missing)

S47FAIL

second register did not surface union aliases; missing ['alphaone-proj', 'a1-project']; by-alias('a1-project') resolved to '', expected 'a944f05f-323c-4100-9c9a-32e91e35feee'; by-alias('alphaone-proj') resolved to '', expected 'a944f05f-...

S48FAIL

check_duplicate returned 0 matches for a near-identical input; original memory (4f67cbd7-ef70-4e01-abda-e255901be664) absent from matches; match did not carry a similarity score field

S49FAIL

promote: HTTP 404

S5PASS

 

S50SKIP

sqlcipher feature not advertised in /api/v1/capabilities — daemon was not built with --features sqlcipher

S51SKIP

autonomous/smart tier not available on node-1 — tier='semantic' models={'cross_encoder': 'none', 'embedding': 'sentence-transformers/all-MiniLM-L6-v2', 'embedding_dim': 384, 'llm': 'none'}

S52FAIL

openclaw_side: signature_verified=False (expected True); openclaw_side: observed_by=None (expected daemon); hermes_side: signature_verified=False (expected True); hermes_side: observed_by=None (expected daemon)

S53FAIL

intruder got 201 (expected 403)

S54PASS

 

S55PASS

 

S56PASS

 

S57PASS

 

S58FAIL

hermes did not see notify with marker=s58-03ff34bdc2d54aad9d6d4a5f344c003f

S59SKIP

v0.7 SSRF guard rejects private-VPC webhook URLs by default; set `[subscriptions] allow_loopback_webhooks = true` on both daemons and export ALLOW_LOOPBACK_WEBHOOKS_VERIFIED=1 to run this scenario.

S6PASS

 

S60FAIL

hermes write to parent got 201 (expected 403)

S61FAIL

openclaw quota.used=0 (expected >=700)

S62PASS

 

S63PASS

 

S64PASS

 

S65FAIL

max_depth=10 found no path A→E

S66PASS

 

S67PASS

 

S68PASS

 

S69PASS

 

S70PASS

 

S71PASS

 

S72PASS

 

S73SKIP

cargo not installed on runner droplet — install rustup or run S73 from a host with cargo. (See docs/coverage.md Postgres+AGE section for the SKIP-condition policy.)

S74SKIP

v0.7.0-alpha ai-memory CLI lacks `probe-capability`/`schema-init`; the SAL UnsupportedCapability error surface is reachable only via cargo-test on the binary. Re-enable when the probe CLI ships.

S75FAIL

pg schema_version=28 (expected 15 for v0.7.0-alpha); missing migration set [] != expected [16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28]

S76PASS

AGE half soft-skipped on s-4vcpu-16gb-amd (cluster has AGE 1.5.0 + memory_graph projection but the bench's cypher-parameter binding hits the AGE 'third argument of cypher function must be a parameter' quirk on PG16+AGE1.5; CTE half measu...

S77PASS

 

S78PASS

 

S79FAIL

10/10 queries returned empty top-K; mean top-5 Jaccard 0.00 < floor 0.80

S80FAIL

hermes write to deep child got 201 (expected 403)

S81PASS

 

S82FAIL

kg_query returned http_code=0; kg_query returned no path between chain endpoints

S9PASS