Gauntlet Vitals as OpenAI Guardrails YAML twin #7
Labels
No labels
burndown-2026-06
coherence-core
consult
headless
interactive
P0
P1
P2
P3
P4
No milestone
No project
No assignees
1 participant
Due date
No due date set.
Dependencies
No dependencies set.
Reference
coilyco-flight-deck/gauntlet#7
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally filed by @coilysiren on 2026-05-03T19:53:36Z - https://github.com/coilysiren/gauntlet/issues/6
Inspiration:
openai/openai-guardrails-python(and the JS variant) - configurable input/output validation for LLM apps, driven by a YAML config.Joins against:
gauntlet's Vitals concept (externally-observable invariants checked independently of the Inspector).Guardrails fire at runtime on every LLM call. Gauntlet's Vitals fire at promotion time on every Trial. The invariants are often the same shape (no PII leaking, no cross-tenant data, expected schema). If gauntlet emits its Vitals as a guardrails YAML twin, the same invariant runs both pre-merge (gauntlet) and at-runtime (guardrails). Two-line value prop. Doesn't require gauntlet to depend on openai-guardrails: just a YAML serializer that produces a config the OpenAI library can consume. Cross-ecosystem credibility on a small surface.
Moved from coilysiren/coilyco-ai#44.
Iceboxed in the 2026-05-29 backlog burn-down: Explore-openai-repo inspiration, speculative future feature. Reopen anytime if it becomes real.