seon ed3d41d399 feat(ai): external responder for external-* model ids (reference runs)
모델 id가 external- 로 시작하면 ai-agent 가 LLM API 를 부르지 않고 각 턴(시스템
프롬프트·대화 이력·도구 정의·강제 제출 여부)을 GET /external/next 로 내어주고
POST /external/reply 로 답을 받는다. 진단 루프·프롬프트·도구·턴 상한은 다른 모델과
동일하므로, API 경로가 없는 모델(Claude Mythos 5.1 을 Claude Code 서브에이전트가
구동)의 참고 측정에 쓴다. 토큰 사용량은 알 수 없어 0 으로 기록한다.

external-* 진단은 EXTERNAL_DIAGNOSE_TIMEOUT_S(기본 1500초), 한 턴은
EXTERNAL_TURN_TIMEOUT_S(기본 1200초)까지 기다리고, 하네스는 같은 모델에 26분을
기다린다. /selfcheck 에 external_responder 상태(대기 턴·마지막 폴링 시각)를 추가.
2026-10-16 09:05:12 +09:00
2026-09-24 01:32:06 +09:00
2026-09-23 23:35:10 +09:00

selfheal-lab-apps

Application sources for the self-healing verification lab (VM + container mixed environment). Infrastructure: seon-labs/selfheal-lab-terraform. Runtime configuration (compose, manifests): selfheal-lab/selfheal-lab-gitops. This repository is public and contains no secrets.

Directory Runs on Image
legacy-api/ GCP MIG instances (docker compose) Artifact Registry asia-northeast1-docker.pkg.dev/cms-d-414903/selfheal/legacy-api:<git sha>
demo-app/ AWS Kubernetes (Argo CD, namespace demo) Public Artifact Registry asia-northeast1-docker.pkg.dev/cms-d-414903/selfheal-public/demo-api:<git sha> (anonymous pull; kubeadm nodes have no cloud credentials)
harness/ GCP harness VM (docker compose) Public Artifact Registry selfheal-public/harness:<git sha>
ai/ GCP AI VM (docker compose): cmd/agent (diagnosis, experiment design 6.2-6.4) and cmd/hitl (Discord approval bot, 6.5-6.6) Public Artifact Registry selfheal-public/ai-agent:<git sha>, selfheal-public/ai-hitl:<git sha> (gcloud builds submit ai --config ai/cloudbuild.yaml --substitutions=_TAG=<git sha>)

Build (Cloud Build, from the repository root):

gcloud builds submit legacy-api --project cms-d-414903 --tag asia-northeast1-docker.pkg.dev/cms-d-414903/selfheal/legacy-api:$(git rev-parse --short HEAD)
S
Description
Self-healing lab application sources (legacy-api for the VM layer, demo-app for Kubernetes). Public, no secrets.
Readme
68 MiB
0 Stars 1 Watchers 0 Forks
Languages
Go 99.2%
Dockerfile 0.8%