Independent deployments will determine whether Substructure’s TOML-and-webhook architecture provides reliable self-hosted durable execution for production agents across programming languages.
state: expiredheat: lowuncertainty: highconvergesscott: mediumagent-harnesses durable-execution self-hostingSubstructure
What is this?
Substructure is presented as a language-agnostic, self-hostable agent engine whose loops are configured with TOML and connected to application code through webhooks rather than a required SDK. Its claimed value is durable agent execution—preserving or reconstructing workflow state after failures—while allowing applications in different languages to participate. The supplied search results explain durable execution and portability concerns generally, but provide no independent deployment evidence validating Substructure’s production reliability; the web summary’s reference to Substructure SDKs also conflicts with the case’s “no SDK” framing.
Why it matters to Scott
Substructure independently implements Scott’s established pattern of durable external state, replaceable language-agnostic workers, and declarative orchestration, directly overlapping his long-running-agent and deterministic control-plane work. Its TOML-and-webhook approach could become a useful production alternative or benchmark, but the lack of independent deployments—and conflicting SDK framing—means it does not yet establish reliability or warrant changing his builds.
ip:framework.long-running-agentsip:concept.durable-external-stateip:framework.12-factor-agents-frameworkdev:concept.deterministic-agent-control-planedev:concept.event-driven-agentsdev:concept.resumable-agent-job-control-planeradar:concept.agent-harnessesradar:concept.long-running-orchestrationradar:hermes-missions-durable-agent-executionradar:pi-agentharness-durable-runtime
queries asked of Scott's wikis
- durable execution for production agent harnesses
- webhook versus SDK agent architecture
- self-hosted agent orchestration and infrastructure sovereignty
- language-agnostic agent runtime design
- crash recovery idempotency and replay for agents
- declarative TOML configuration for agent loops
Measured heat
no measured readings yet — the hourly heat pass fills this in
How the heat travelled
no chain yet — the hourly chain pass fills this in
Evidence (2) — ⭐ canonical anchor
Interpretation history
2026-08-15T15:30:02Z
The launch failed to attract discussion, independent implementations, or deployment evidence within its initial observation window. Substructure remains a potentially relevant architecture, but this episode has faded without validating its production-reliability hypothesis.
2026-08-13T14:42:42Z
The reobservation adds only negligible engagement and no independent deployment, implementation, or reliability evidence. The case remains an unvalidated first-party architecture launch awaiting real-world use.
2026-08-13T14:29:51Z
grounded: converges/medium — Substructure independently implements Scott’s established pattern of durable external state, replaceable language-agnostic workers, and declarative orchestratio
2026-08-13T14:27:16Z
case created — The first-party launch exposes a concrete agent-runtime artifact, although it has not yet attracted independent deployment evidence.
Decision trace
- 08-16 01:30expireThe launch failed to attract discussion, independent implementations, or deployment evidence within its initial observation window. Substructure remains a potentially relevant architecture, but this e
- 08-16 01:30alert_silentThe only change is negligible engagement with no comments or substantive evidence; there is no new event that Scott needs before a future concrete deployment or technical validation emerges.
- 08-16 01:30alert_routeThe only change is negligible engagement with no comments or substantive evidence; there is no new event that Scott needs before a future concrete deployment or technical validation emerges.
- 08-14 00:42repriceThe reobservation adds only negligible engagement and no independent deployment, implementation, or reliability evidence. The case remains an unvalidated first-party architecture launch awaiting real-
- 08-14 00:42alert_silentNo material event occurred beyond a one-point engagement increase; the architecture and durability claims remain unchanged and can wait for concrete deployment evidence.
- 08-14 00:42alert_routeNo material event occurred beyond a one-point engagement increase; the architecture and durability claims remain unchanged and can wait for concrete deployment evidence.
- 08-14 00:37alert_silentThis is a first-party public introduction of a relevant architecture, but the durability and production-reliability claims remain unvalidated, with no independent deployments, technical artifacts, or
- 08-14 00:37surface_candidateThis is a first-party public introduction of a relevant architecture, but the durability and production-reliability claims remain unvalidated, with no independent deployments, technical artifacts, or
- 08-14 00:37alert_routeThis is a first-party public introduction of a relevant architecture, but the durability and production-reliability claims remain unvalidated, with no independent deployments, technical artifacts, or
- 08-14 00:29groundSubstructure independently implements Scott’s established pattern of durable external state, replaceable language-agnostic workers, and declarative orchestration, directly overlapping his long-running
- 08-14 00:27createThe first-party launch exposes a concrete agent-runtime artifact, although it has not yet attracted independent deployment evidence.