Evenfire is an intelligence layer platform for running LLM agents that take real actions, on your infrastructure
Evenfire repository guidance
Shared capability naming
Name shared variables, functions, types, database fields, quota families, metrics and configuration after the domain or capability they implement. Do not name shared infrastructure after the AI agent, authoring tool, model or provider that first introduced it. A shared subscription quota, for example, uses an administrative subscription name.
Provider names remain appropriate for actual provider-specific adapters, protocol identifiers and external contracts. Historical data and migration compatibility identifiers may retain legacy names when the concrete compatibility requirement is documented. Keep those references confined to that boundary; do not copy them into new shared runtime names.
Branch naming
Never create new branches with agent or vendor prefixes such as codex/*,
claude/*, openai/*, anthropic/*, antrophic/*, or any other Frontier
Labs reference. Use the repository's conventional prefixes instead: feat/*,
fix/*, hotfix/*, chore/*, docs/*, test/*, refactor/*, ci/*,
build/*, perf/*, or revert/*.
The local Minikube T0/T1/T2 contract is development-only. Follow the
canonical make minikube-t2-preflight, make minikube-t2, and
make minikube-t2-runtime entry points and the ownership, secret-safety, and
evidence rules in AGENTS.md and docs/testing/minikube-t2-runbook.md.
Hold Control UI / Desktop port-forwards on the host with
make -f scripts/minikube-profiles/branch.mk branch-profile-pf
(AGENTS.md § Branch-profile UI port-forwards). Do not use
make minikube-pf-all-bg as the hold.
make minikube-t2-preflight is a planner, not a lane verdict. A T2 verdict is
produced only by the final exact-head preflight inside make minikube-t2 (or
make minikube-t2-runtime after T0 and T1 are already green on the same HEAD).
make minikube-pre-gate-sync alone is not T2. Playwright and product E2E
scripts are separate lanes. T1 requires both a complete green JSON reporter
(expected files, executed/passed, zero failures, zero pending) and a zero
Vitest process exit; a green reporter cannot hide teardown, worker, OOM, or
signal failure. The single run is
self-healing: the orchestrator planner selects full-reconcile for a
bootstrapped profile with an unready deployment (never PROFILE_UNHEALTHY
before a transition), T1 restores branch-profile GFS credentials on exit, and
pre-gate-sync provisions GFS serving with GFS_RESTORE_ACTIVE_NOLOGIN=true
and GFS_RECOVER_ABANDONED_STATE=true only in the minikube-t2 transition;
other security gates refresh MCP auth without mutating GFS. Do not insert
manual repair scripts between runs. Harness GFS reconciles settle Ready-reader
leftovers first (settle-gfs-reader-rollout.sh) and judge reader rollout
completion by readiness via the gfs-rollout-shim PATH prefix, because
HCC's gfsReconciler now preserves the restartedAt annotation, and the
harness still judges reader readiness via the shim so leftover ReplicaSets
cannot turn a generation-based kubectl rollout status into a timeout.
Follow the rule
.cursor/rules/minikube-t0-t1-t2.mdc and the skill
.cursor/skills/minikube-t0-t1-t2/SKILL.md for the certification workflow.
Logging standard
For production Node.js/runtime code, use structured logging. Pino is the
repository reference implementation because control-api uses a redacting
root logger and workspace-files-controller uses Pino/Pino HTTP; Winston is
not a repository dependency. Services with an existing structured wrapper
(mcp-host, HCC, workflow recipes, and the workflow approval reader) should
continue using that wrapper rather than bypassing it.
When editing a file, replace its direct console.* calls with the existing
service logger, preserve levels/context/redaction, and keep the change scoped
to that file plus minimal support. Do not start a repository-wide logging
migration or add a second logger convention. Direct console output is reserved
for logger adapters, explicit CLI/bootstrap output, tests, E2E/fixtures, and
intentional development diagnostics; never emit secrets, tokens, credentials,
DSNs, private URLs, or raw request/response bodies.
Desktop Electron runtime
A skipped Electron postinstall leaves the runtime undownloaded while npm ci
still exits successfully, so the failure surfaces later as a test error that
looks like a product bug. No package.json field prevents this; the guard is
npm run verify:electron, which fails when require('electron') does not
resolve to an executable. Desktop validation must use Node 24, run npm ci
without --ignore-scripts, and pass that check before any test/build result is
counted. If the check reports Electron failed to install correctly, switch to
Node 24, repair the generated dependency directory, and rerun the install.
Never bypass the postinstall with ELECTRON_SKIP_BINARY_DOWNLOAD, use an
unverified override, or treat the resulting test failure as a product
regression.