RunRecord.result and Tropibot completion outcome.result use schemaVersion: 1. Enum values are stable snake_case strings. Unknown future result fields should be ignored by consumers; request version 1 is intentionally strict and rejects unknown fields.
Top-level fields:
| Field | Meaning |
|---|---|
schemaVersion |
Result contract version, currently 1. |
runId |
Caller-selected safe idempotency key. |
source |
Repository, pull request number, and head SHA supplied by the trusted build flow. |
package |
Target name, requested references, and versions reported by Dappmanager. |
execution |
Completed execution status and UTC timing. Execution status is separate from verdict. |
verdict |
passed, failed, warning, inconclusive, or infrastructure_error. |
reasonCode |
Central machine-readable reason such as candidate_containers_stable. |
summary |
Bounded human-readable explanation. |
baseline, candidate |
Install timing, hard check, normalized containers, and log collection metadata. |
comparison |
Added/removed containers, versions, timing, last non-running states, and deterministic regressions. |
logAnalysis |
Advisory heuristic, Nexus, or composite analysis with bounded findings. |
cleanup |
Cleanup status, bounded error, and packages not present in the initial snapshot. |
errors |
Bounded typed run errors with the phase where each occurred. |
An unresolvable published baseline completes with verdict: "inconclusive", reasonCode: "baseline_unresolvable", and cleanup.status: "skipped". Its summary explains that upgrade comparison was skipped without changing packages and suggests repairing or publishing the registry entry or supplying a known-good IPFS baselineRef. Both sides' install and log-collection statuses remain pending to mean not attempted; their hard checks are false with reason baseline_unresolvable, rather than falsely reporting unstable containers. The coordinator receives and acknowledges this normal result, and the worker can accept the next job. Generic RPC failures and contract reverts are not treated as evidence of a missing baseline.
The local recovery plan can be restore_baseline for an initially installed target with an explicit IPFS baseline. Its immutable reference comes from the persisted request's package.baselineRef, so recovery does not require a published release. Cleanup preserves volumes and verifies the restored package's IPFS origin. Existing records with restore_latest or remove retain their recorded recovery policy.
The v1 coordinator container shape remains name, serviceName, state, running, image, and created. Optional Docker health, container identity, restart count, and start time are kept only in local capture diagnostics, along with each stabilization sample's safe probeResults. No probe URL, authentication header, or response body is persisted. Existing reason codes such as candidate_containers_stable and candidate_containers_unstable now include configured application readiness and reported Docker health in the hard check. Cleanup passes only after restoration identity/readiness and final inventory verification; it fails for changed or missing unrelated packages as well as unexpected additions.
If Dappmanager rejects the baseline package signature, the harness retries that
baseline mutation with the signed-package restriction bypass. A successful
retry continues the comparison, changes an otherwise passing verdict to
warning with reason code baseline_signature_invalid, and preserves the
original rejection in errors.
Representative successful result:
{
"schemaVersion": 1,
"runId": "pr-123-abcdef0-example-package",
"source": {
"repository": "dappnode/example-package",
"pullRequest": 123,
"headSha": "abcdef0123456789"
},
"package": {
"dnpName": "example.dnp.dappnode.eth",
"baselineRequestedRef": null,
"baselineResolvedVersion": "1.0.0",
"candidateRef": "/ipfs/QmCandidate",
"candidateReportedVersion": "1.0.1"
},
"execution": {
"status": "completed",
"startedAt": "2026-07-10T12:00:00Z",
"finishedAt": "2026-07-10T12:01:00Z",
"durationMs": 60000
},
"verdict": "passed",
"reasonCode": "candidate_containers_stable",
"summary": "Baseline and candidate passed readiness checks",
"baseline": {
"install": { "status": "passed", "durationMs": 1000 },
"hardCheck": {
"passed": true,
"reasonCodes": [],
"containerCount": 1,
"stableSamples": 3
},
"containers": [],
"logCollection": { "status": "passed", "containerCount": 1 }
},
"candidate": {
"install": { "status": "passed", "durationMs": 1000 },
"hardCheck": {
"passed": true,
"reasonCodes": [],
"containerCount": 1,
"stableSamples": 3
},
"containers": [],
"logCollection": { "status": "passed", "containerCount": 1 }
},
"comparison": {
"baselineHardCheck": true,
"candidateHardCheck": true,
"baselineContainers": ["service"],
"candidateContainers": ["service"],
"containersAdded": [],
"containersRemoved": [],
"baselineVersion": "1.0.0",
"candidateVersion": "1.0.1",
"baselineStabilizationMs": 10000,
"candidateStabilizationMs": 10000,
"baselineLastNonRunningStates": [],
"candidateLastNonRunningStates": [],
"baselineLogsCollected": true,
"candidateLogsCollected": true,
"deterministicRegressions": []
},
"logAnalysis": {
"analyzer": "heuristic",
"status": "clean",
"summary": "No configured candidate-only suspicious signature was found",
"baseline": { "status": "clean", "summary": "No configured suspicious signature was found" },
"candidate": { "status": "clean", "summary": "No configured suspicious signature was found" },
"newFindings": []
},
"cleanup": { "status": "passed", "leftoverPackages": [], "error": null },
"errors": []
}Tropibot receives a normal result in this envelope:
{
"schemaVersion": 1,
"workerId": "worker-01",
"claimToken": "opaque-random-value",
"outcome": {
"type": "result",
"result": {}
}
}If no normal result can be safely produced, the worker sends type: "worker_error" with one of interrupted, unsupported_job, cleanup_failed, local_persistence_failed, or unexpected_error, plus a bounded summary and cleanupStatus. The exact serialized envelope is stored before delivery and retried unchanged until Tropibot replies with recorded or duplicate.
The persisted surrounding RunRecord includes request, phase history, bounded redacted capture evidence, local worker-delivery state, and this result. The opaque claim token must be retained locally to complete after restart; the Tropibot shared bearer token, Dappmanager token, Nexus key, authorization headers, and raw unredacted logs are never result-schema fields.