convez.ai
SkillDevOps & InfraestruturaAvançado
chf3198

Megingjord Harness

chf3198/megingjord-harness

Governança unificada para agentes IA: deploy, roteamento e compliance

Ver no GitHub
InicialPluginScore · de 0 a 100

Plataforma de governança para agentes de IA que orquestra Copilot, Claude Code e Codex através de um modelo de baton workflow, roteamento inteligente de LLMs (Ollama/Claude/OpenRouter) e gates de CI automatizados. Funciona como um control plane que deploy skills, hooks, instruções e políticas de forma sincronizada em múltiplos runtimes e projetos.

Instalação

$npx convez-ai chf3198/megingjord-harness

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
1
Acompanhando
0
Saúde
100%
Último commit
há 1 dia
Versão
v3.4.1
Issues abertas
201
Contribuidores
2

Casos de uso

  • Governança centralizada de múltiplos agentes IA em equipes
  • Deploy sincronizado de skills e políticas entre Copilot, Claude Code e Codex
  • Implementação de gates de CI/CD e telemetria de uso de agentes
  • Reutilização de conhecimento institucional via LLM wiki
  • Controle de acesso e auditoria em fleets de agentes

Veredito

Solução madura e bem arquitetada para equipes que precisam governar múltiplos agentes IA em produção. Requer expertise em Node.js e entendimento de workflows de CI/CD, mas oferece controle granular e auditoria robusta — excelente para contextos empresariais com conformidade rigorosa.

README

A tradução para o português deste README ainda está na fila.

Megingjord Harness

Megingjord Banner

License: PolyForm NC Node >=22 Plugin Runtime Quality Gates OpenSSF Scorecard

Megingjord is a governance-first AI agent harness with skills, hooks, agents, runtime scripts, and a Karpathy-style LLM wiki. It operates as a control plane for AI coding agent runtimes — governing Copilot, Claude Code, and Codex the way Kubernetes governs containers: deploying configuration, enforcing policy, and maintaining desired workflow state across all three simultaneously.

Architecture at a glance

flowchart LR
	A[Repo Source] --> B[skills/ instructions/ hooks/ scripts]
	B --> C[~/.copilot runtime]
	B --> D[~/.codex/devenv-ops runtime]
	B --> E[~/.agents/skills runtime]
	C --> F[Governed agent execution]
	D --> F
	E --> F

Why it is robust

  • Multi-runtime deployment model with dry-run/apply scripts
  • Governance baton model: Manager → Collaborator → Admin → Consultant
  • Fleet-aware routing, telemetry, and policy enforcement
  • Confidence-aware telemetry separates exact vs estimated Copilot usage with report caveats
  • Static dashboard with operations + governance visibility
  • LLM wiki integration for reusable institutional knowledge
  • Accessibility + UX baseline checks aligned to WCAG 2.2 and Core Web Vitals

How it works — two-tier model

Installing Megingjord in a project seeds two independent layers:

Global layer — lives in your home directory, shared by all projects:

RuntimePathWhat lives there
GitHub Copilot Chat~/.copilot/skills, instructions, hooks, scripts, wiki
Claude Code~/.claude/commands, agents, hooks, settings
Codex~/.codex/AGENTS.md, config, hooks, rules

One npm run deploy:apply and every project on the machine inherits the full governance toolkit — skills, routing, telemetry, wiki — for all three runtimes.

Workspace layer — checked into each project repo, workspace-specific:

ExtensionFile(s)Purpose
Copilot.github/copilot-instructions.mdworkspace override + adapter
Claude CodeCLAUDE.md, .claude/settings.jsonworkspace override + adapter
CodexAGENTS.md, .codex/workspace override + adapter

Workspace files extend or override the global layer. Global governance wins by default; local files add project-specific context and commands.

Multi-project reuse: npm run deploy:apply is idempotent. Multiple projects share one global skill and governance layer while each keeps its own baton history, workspace wiki, and local overrides.

Multi-runtime parity: A skill, hook, or governance rule written once deploys to all three runtimes — Copilot, Claude Code, and Codex are equal first-class citizens.

Zero-Cloudflare by default: Cross-agent coordination (mailbox, bundles, telemetry, async dispatch) uses GitHub-native primitives out of the box — no Cloudflare account or wrangler deploy required. Set MEGINGJORD_HAMR_ENABLED=1 to opt into the accelerated HAMR Cloudflare Worker for higher-throughput paths. See docs/howto/github-native-layer2.md.

See docs/howto/installation.md for the full install walkthrough, including adding a second project.

Ownership Semantics Rollout

Execution role labels represent the active baton holder only. Waiting and terminal states must not carry execution role labels.

See docs/howto/baton-workflow.md for board filters, no-code remediation lane constraints, and the operator rollout checklist.

Quick start

npm run setup
npm start
npm run lint
npm test
npm run deploy:both:apply

No Cloudflare account needed. Cross-agent coordination works immediately via GitHub-native primitives. To enable the HAMR accelerated path after deploying the Worker: add MEGINGJORD_HAMR_ENABLED=1 to your .env.

Env-backed credentials

  • Copy .env.example to .env for local resource credentials.
  • Tavily MCP must keep the secret only in TAVILY_API_KEY; do not inline Tavily keys in MCP URLs or tracked config.
  • Verify env-backed Tavily wiring with npm run governance:credentials-env.

Research-first gate notes

  • Research-first Epic detection is label-first: phase-gate:research-first on type:epic tickets.
  • AC-R* body matching remains a legacy fallback during migration and is slated for removal after #1888 enforcement follow-on completes.
  • Phase-1 MANAGER_HANDOFF checks use phase-gate:phase-1 by default and can be overridden with PHASE_ONE_LABEL for validation runs.

Scripts

ScriptCommand
adr:buildlog4brains build
adr:newlog4brains adr new
adr:previewlog4brains preview
agent:coord:remotenode scripts/global/agent-coord-remote.js
agent:tier-cnode scripts/global/tier-c-guard.js
anneal:detectnode scripts/global/workflow-anneal-detect.js
anneal:goal-sensornode scripts/global/anneal-goal-sensor.js
anneal:graphnode scripts/global/anneal-graph.js
anneal:reviewnode scripts/global/anneal-review.js
anneal:rotatenode scripts/global/anneal-log-rotate.js
anneal:schedule-healthnode scripts/global/anneal-schedule-health.js
auth:auditnode scripts/global/authorization-audit.js
auth:profilenode scripts/global/authorization-profile.js
auth:profile:contextnode scripts/global/authorization-profile-context.js
capability:probenode scripts/global/capability-probe.js
capability:shownode scripts/global/capability-show.js
catch-emptybash scripts/global/catch-empty-lint.sh .github/workflows
cleanup:branchesnode scripts/global/branch-cleanup-plan.js
collaborator:preflightnode scripts/global/collaborator-preflight.js
cost-reportnode scripts/global/cost-report.js
cost:baselinenode scripts/global/cost-baseline.js
cost:token-reportnode scripts/global/token-spend-report.js
deploybash scripts/deploy.sh && node scripts/global/xteam-mcp-register.js --target copilot --root .
deploy:allbash scripts/deploy.sh --target all && node scripts/global/xteam-mcp-register.js --target all --root .
deploy:all:applybash scripts/deploy.sh --apply --target all && node scripts/global/xteam-mcp-register.js --target all --root . --apply
deploy:antigravitybash scripts/deploy.sh --target antigravity && node scripts/global/xteam-mcp-register.js --target antigravity --root .
deploy:antigravity:applybash scripts/deploy.sh --apply --target antigravity && node scripts/global/xteam-mcp-register.js --target antigravity --root . --apply
deploy:applybash scripts/deploy.sh --apply && node scripts/global/xteam-mcp-register.js --target copilot --root . --apply
deploy:bothbash scripts/deploy.sh --target both && node scripts/global/xteam-mcp-register.js --target copilot --root . && node scripts/global/xteam-mcp-register.js --target codex --root .
deploy:both:applybash scripts/deploy.sh --apply --target both && node scripts/global/xteam-mcp-register.js --target copilot --root . --apply && node scripts/global/xteam-mcp-register.js --target codex --root . --apply
deploy:claudebash scripts/deploy.sh --target claude && node scripts/global/xteam-mcp-register.js --target claude --root .
deploy:claude:applybash scripts/deploy.sh --apply --target claude && node scripts/global/xteam-mcp-register.js --target claude --root . --apply
deploy:codexbash scripts/deploy.sh --target codex && node scripts/global/xteam-mcp-register.js --target codex --root .
deploy:codex:applybash scripts/deploy.sh --apply --target codex && node scripts/global/xteam-mcp-register.js --target codex --root . --apply
deps:aggregatenode scripts/global/dep-graph-aggregate.js
deps:augmentnode scripts/global/dep-graph-augment.js
deps:rendernode scripts/global/dep-graph-render.js
deps:reviewnode scripts/global/dep-proposals-review.js
doc-graphnode scripts/global/doc-graph-builder.js
docs:anchorsnode scripts/global/docs-anchors.js
docs:compilenode scripts/docs-compile.js
docs:execnode scripts/global/docs-exec.js
docs:lintnode scripts/docs-lint.js
epic:syncnode scripts/global/actuator-epic-sync.js
formatprettier --write .prettierrc.json package.json CONTRIBUTING.md .github/workflows/lint.yml lint-configs/README.md lint-configs/ci-lint.yml lint-configs/eslint.config.devenv.js scripts/lint-readability.js scripts/global/install-readability-toolchain.js
format:checkprettier --check .prettierrc.json package.json CONTRIBUTING.md .github/workflows/lint.yml lint-configs/README.md lint-configs/ci-lint.yml lint-configs/eslint.config.devenv.js scripts/lint-readability.js scripts/global/install-readability-toolchain.js
git-state:driftnode scripts/global/git-state-drift-sensor.js
git:conflict-predictnode scripts/global/git-conflict-predict.js
git:conflict-predict:testnode --test tests/git-conflict-predict.spec.js
git:freshness-checknode scripts/global/git-freshness-check.js
git:freshness-check:testnode --test tests/git-freshness-check.spec.js
goals:regennode scripts/global/goals-contract-generate.js
gov:extractnode scripts/global/rule-card-extractor.js --all > /tmp/rule-cards.json
gov:paritynode scripts/global/megalint/parity-validator.js
governance:actuator-transitions:testnode --test tests/actuator-transitions.spec.js
governance:adapters:emitnode scripts/global/governance-adapter-emit.js
governance:adapters:testnode scripts/global/governance-adapter-emit.spec.js
governance:anneal-decision-auditnode scripts/global/anneal-decision-audit.js
governance:anneal-decision-detector:testnode --test tests/anneal-decision-detector.spec.js
governance:auditnode scripts/global/governance-audit.js
governance:baton-latency:testnode --test tests/baton-latency-report.spec.js
governance:collab-handoff-rebase:testnode --test tests/collab-handoff-rebase-freshness.spec.js
governance:compatibility:matrixnode scripts/global/governance-compatibility-matrix.js
governance:coordinator-cleanupnode scripts/global/coordinator-label-cleanup.js
governance:credentials-envnode scripts/global/credentials-env-guard.js
governance:cross-team-checknode scripts/global/cross-team-contract-check.js
governance:cross-team-check:testnode --test tests/cross-team-contract-check.spec.js
governance:cross-team-lease-dispatch:testnode --test tests/cross-team-lease-dispatch.spec.js
governance:delegation-lintnode scripts/global/delegation-phrase-lint.js
governance:dispatcher-set-labels:testnode --test tests/github-dispatcher-set-labels.spec.js
governance:driftnode scripts/global/governance-drift-classifier.js --json
governance:duplicate-checknode scripts/global/ticket-duplicate-guard.js --scan
governance:duplicate-check:testnode --test tests/ticket-duplicate-guard.spec.js
governance:epicnode scripts/global/epic-evidence.js
governance:epic-auto-tick:testnode --test tests/epic-task-list-auto-tick.spec.js
governance:epicsnode scripts/global/epic-close-validator.js
governance:escalation-coveragenode scripts/global/escalation-coverage-gate.js
governance:escalation-coverage:testnode --test tests/escalation-coverage-gate.spec.js
governance:generatenode scripts/global/governance-generate.js
governance:ghs-feedback-loop:testnode --test tests/ghs-feedback-loop.spec.js
governance:goal-tier-resolver:testpython3 -m unittest tests.hooks.test_goal_tier_resolver
governance:hamr-bypass-workflow:testnode --test tests/hamr-bypass-lint-workflow.spec.js
governance:hook-paritynode scripts/global/hook-parity-check.js
governance:hook-parity:testnode --test tests/hook-parity-check.spec.js
governance:hooks:testpython3 -m unittest discover -s tests/hooks -p 'test_*.py'
governance:instructional-coveragenode scripts/global/instructional-coverage-audit.js
governance:instructional-coverage:testnode --test tests/instructional-coverage-audit.spec.js
governance:issue-transition-dispatch:testnode --test tests/issue-transition-dispatch.spec.js
governance:lease-heartbeatnode scripts/global/worktree-lease-heartbeat.js
governance:lease-heartbeat:testnode --test tests/worktree-lease-heartbeat.spec.js
governance:list-ticketsnode scripts/global/ticket-presenter.js
governance:list-tickets:testnode --test tests/ticket-presenter.spec.js
governance:load-local-env:testnpx playwright test tests/load-local-env.spec.js
governance:manifest:testnode scripts/global/governance-manifest-validate.spec.js
governance:manifest:validatenode scripts/global/governance-manifest-validate.js
governance:model-diversity-replaynode scripts/global/model-diversity-replay-eval.js
governance:model-diversity-replay:testnode --test tests/model-diversity-replay-eval.spec.js
governance:no-sync-httpnode scripts/global/no-sync-http-handlers.js
governance:operator-ownership:evalnode scripts/global/operator-ownership-eval.js
governance:operator-ownership:testnode --test tests/operator-ownership-rules.spec.js tests/delegation-phrase-lint.spec.js tests/operator-ownership-eval.spec.js tests/runtime-side-effect-guard.spec.js
governance:orchestrator-paritynode scripts/global/orchestrator-governance-parity.js --json
governance:orchestrator-parity:testnode --test tests/orchestrator-governance-parity.spec.js
governance:parity:testnode scripts/global/governance-parity.spec.js
governance:pre-commit-docs-checknode scripts/global/pre-commit-docs-check.js
governance:pre-commit-docs-check:testnode --test tests/pre-commit-docs-check.spec.js
governance:pre-pr-gatenode scripts/global/pre-pr-gate.js
governance:pre-pr-gate:testnode --test tests/pre-pr-gate.spec.js
governance:price-capnode scripts/global/provider-price-cap-gate.js
governance:price-cap:testnode --test tests/provider-price-cap-gate.spec.js
governance:reconcilenode scripts/global/ticket-reconcile.js --json
governance:review-promotion-evalnode scripts/global/model-diversity-replay-eval.js && node scripts/global/second-opinion-replay-eval.js
governance:review-scorenode scripts/global/review-score-classifier.js
governance:review-score:testnode --test tests/review-score-classifier.spec.js
governance:role-baton-auditnode scripts/global/role-baton-audit.js
governance:role-baton-linternode scripts/global/role-baton-linter.js
governance:role-baton-linter:testnode --test tests/role-baton-linter.spec.js
governance:rule-paritynode scripts/global/governance-rule-parity.js
governance:rule-parity:testnode --test tests/governance-rule-parity.spec.js
governance:runtime-guard:testnode --test tests/runtime-side-effect-guard.spec.js
governance:second-opinion-replaynode scripts/global/second-opinion-replay-eval.js
governance:second-opinion-replay:testnode --test tests/second-opinion-replay-eval.spec.js
governance:second-opinion-runnernode scripts/global/second-opinion-runner.js
governance:second-opinion-runner:testnode --test tests/second-opinion-runner.spec.js
governance:second-opinion-tier3:testnode --test tests/second-opinion-tier3-trigger.spec.js
governance:soak-languagenode scripts/global/megalint/soak-language-guard.js
governance:soak-language:testnode --test tests/soak-language-guard.spec.js
governance:status-cardinalitynode scripts/global/label-lint-status-cardinality.js
governance:status-cardinality:testnode --test tests/label-lint-status-cardinality.spec.js
governance:sync-checknode scripts/global/governance-sync-check.js
governance:ticket-redundancy:testnode --test tests/validate-ticket-redundancy.spec.js
governance:tokensnode scripts/global/governance-token-lint.js
governance:validate-config-schemas-workflow:testnode --test tests/validate-config-schemas-workflow.spec.js
governance:verifynode scripts/global/governance-verify.js --json
governance:weeklynode scripts/global/governance-weekly-report.js
governance:wiki-catalog:testnode --test tests/validate-wiki-catalog.spec.js
governance:wiki-metrics:testnode --test tests/wiki-metrics.spec.js
governance:worktree-intercept:testnode --test tests/worktree-write-intercept.spec.js
governance:worktree-inventorynode scripts/global/worktree-inventory.js
governance:worktree-lock:testnode --test tests/worktree-active-session-lock.spec.js
governance:worktreesnode scripts/global/worktree-governance-audit.js --json
hamr:activatebash scripts/global/hamr-activate.sh
hamr:batch-route:testnode --test tests/batch-route-async.spec.js
hamr:batch-routernode scripts/global/anthropic-batch-router.js
hamr:batch-validatenode scripts/global/batch-validator.js
hamr:cache-emitnode scripts/global/cache-stats-emit.js
hamr:cache-gatenode scripts/global/cache-hit-gate.js
hamr:cache-gate:testnode --test tests/cache-hit-gate-eligibility.spec.js
hamr:cache-pushnode scripts/global/cache-stats-push.js
hamr:compressnode scripts/global/constitution-compressor.js
hamr:deploybash scripts/global/hamr-deploy.sh
hamr:doctornode scripts/global/hamr-doctor.js
hamr:fleet-cascade-gatenode scripts/global/fleet-cascade-gate.js
hamr:fleet-cascade-gate:testnode --test tests/fleet-cascade-gate.spec.js
hamr:governance-bundle-pushnode scripts/global/governance-bundle-push.js
hamr:healthnode scripts/global/substrate-health.js
hamr:health-pushnode scripts/global/substrate-health-push.js
hamr:install-cronbash scripts/global/install-cron.sh
hamr:log-rotatenode scripts/global/log-rotate.js
hamr:multi-judge:testnode --test tests/multi-judge-prompts.spec.js tests/multi-judge-variance.spec.js tests/multi-judge-orchestrator.spec.js
hamr:policy-overridesnode scripts/global/cascade-policy-overrides.js
hamr:premium-budget:testnode --test tests/premium-budget-governor.spec.js
hamr:rule-gatenode scripts/global/rule-coverage-gate.js
hamr:skill-viewsnode scripts/global/skill-views-derive.js
hamr:spillovernode scripts/global/header-spillover.js
hamr:sticky-routenode scripts/global/sticky-route.js
hamr:sync-verifynode scripts/global/hamr-sync-verify.js
hamr:teardownbash scripts/global/hamr-teardown.sh
harness:self-testnode scripts/global/harness-self-test.js
harness:self-test:capabilitynode scripts/global/harness-self-test.js --capability
harness:self-test:regressionnode scripts/global/harness-self-test.js --regression
harness:self-test:testnode --test tests/harness-self-test-runner.spec.js tests/harness-self-test-reporters.spec.js tests/harness-self-test-telemetry.spec.js tests/harness-self-test.spec.js
healthnode scripts/health-check.js
help:topicnode scripts/help-topic.js
hooks:installbash scripts/global/install-hooks.sh
hooks:pre-pushnode scripts/global/pre-push-gates.js
issue:transitionnode scripts/global/issue-transition.js
it-ops:usage-reportnode scripts/global/it-bypass-usage-report.js
label:provisionnode scripts/global/label-provision.js
lintnode scripts/lint.js
lint:allnpm run lint:js && npm run lint:py && npm run lint:sh && npm run lint:md && npm run lint:decisions
lint:coverage-metricnode scripts/global/lint-coverage-metric.js
lint:decisionsnode scripts/global/decisions-md-validator.js
lint:jseslint -c lint-configs/eslint.config.devenv.js --max-warnings 9999 dashboard/js scripts/global scripts/wiki
lint:mdmarkdownlint-cli2 '**/*.md' '!node_modules/**' '!scripts/xteam-mcp/node_modules/**' '!.claude/**' '!research/**' '!wiki/wisdom/global/sources/**' '!wiki/wisdom/global/syntheses/**' '!raw/**' '!.dashboard/**' '!CHANGELOG-archive.md' '!tickets/**' '!planning/**' '!generated/**' '!tests/fixtures/governance/golden/**' '!tests/fixtures/changelog-fragments-invalid/**'
lint:pyruff check --config lint-configs/ruff.devenv.toml hooks/scripts/
lint:readabilitynode scripts/lint-readability.js
lint:readability:cinode scripts/lint-readability.js --max-warnings=475
lint:routernode scripts/lint-router.js
lint:shfind scripts -name '*.sh' -exec shellcheck --severity=warning {} +
mailbox:flushnode scripts/global/mailbox-outbox.js flush
mailbox:pollnode scripts/global/mailbox-client.js poll
mailbox:sendnode scripts/global/mailbox-client.js send
mcp:project-statenode scripts/global/mcp-project-state.js
mcp:registernode scripts/global/xteam-mcp-register.js --target all --root . --apply
merge-evidence:snapshotnode scripts/global/merge-evidence-snapshot.js
preparebash scripts/install-git-hooks.sh
quality:paritynode scripts/global/quality-parity-report.js --json
rag:searchnode scripts/global/rag-search.js
readability:snapshotbash scripts/readability-snapshot.sh
repo:scopenode scripts/global/repo-scope.js
rotation:checknode scripts/global/hamr-rotation-check.js
router:cascadenode scripts/global/cascade-dispatch.js
router:dispatchnode scripts/global/task-router-dispatch.js
router:freenode scripts/global/free-router.js
router:smokenode scripts/global/task-router-smoke.js
router:weeklynode scripts/global/model-routing-weekly-report.js
routing:baselinenode scripts/global/routing-baseline-report.js
routing:calibratenode scripts/global/cascade-calibrate.js
routing:fallback-reportnode scripts/global/routing-fallback-report.js
routing:fleet-cost-reportnode scripts/global/fleet-dev-cost-report.js
routing:free-cloud-reportnode scripts/global/free-cloud-usage-report.js
routing:freshnessnode scripts/global/matrix-freshness.js
routing:reconcilenode scripts/global/token-telemetry-reconcile.js --json
routing:refreshnode scripts/global/routing-refresh.js --update-matrix
routing:reportnode scripts/global/routing-baseline-report.js --days 7
routing:telemetrynode scripts/global/token-telemetry-report.js --json
setupnpm install && echo '✅ megingjord ready — run: npm start'
signing:bootstrapnode scripts/global/crypto-signing-bootstrap.js
startnode scripts/dashboard-server.js
state:offloadnode scripts/global/state-offload-client.js
stressMEGINGJORD_STRESS_TIER=A node scripts/global/stress-orchestrator.js
stress:annealnode --test tests/stress-anneal-decision.spec.js
stress:boundedMEGINGJORD_STRESS_TIER=C node scripts/global/stress-orchestrator.js
stress:evidence-checknode scripts/global/stress-evidence-check.js
stress:evidence-check:testnode --test tests/stress-evidence-check.spec.js
stress:fullMEGINGJORD_STRESS_TIER=D node scripts/global/stress-orchestrator.js
stress:load-local-envnode --test tests/stress-load-local-env.spec.js
stress:post-merge-sweepnode --test tests/stress-post-merge-sweep.spec.js
stress:realismMEGINGJORD_STRESS_TIER=B node scripts/global/stress-orchestrator.js
stress:rebasenode --test tests/stress-rebase-discipline.spec.js
stress:role-batonnode --test tests/stress-role-baton-linter.spec.js
stress:surface-auditnode scripts/global/stress-surface-audit.js
stress:surface-audit:testnode --test tests/stress-surface-audit.spec.js
stress:testnode scripts/global/stress-runner.js
stress:wiki-metrics:testnode --test tests/stress-wiki-metrics.spec.js
stress:worktreenode --test tests/stress-worktree-isolation.spec.js
syncbash scripts/sync.sh
sync:allbash scripts/sync.sh --target all
sync:all:drybash scripts/sync.sh --dry-run --target all
sync:bothbash scripts/sync.sh --target both
sync:both:drybash scripts/sync.sh --dry-run --target both
sync:claudebash scripts/sync.sh --target claude
sync:claude:drybash scripts/sync.sh --dry-run --target claude
sync:codexbash scripts/sync.sh --target codex
sync:codex:drybash scripts/sync.sh --dry-run --target codex
sync:drybash scripts/sync.sh --dry-run
synthesis:initnode scripts/global/synthesis-init.js
synthesis:snapshotnode scripts/global/synthesis-snapshot.js
synthesis:statusnode scripts/global/broker-synthesis-status.js
testnpx playwright test
test:collaborator-input-shapenode --test tests/collaborator-handoff-input-shape.spec.js
test:compatibilitynode --test tests/orchestrator-compatibility.spec.js
test:headednpx playwright test --headed
test:post-merge-sweepnode --test tests/post-merge-sweep.spec.js
test:qualitynpx playwright test tests/google-quality.spec.js
test:sse-loadnode scripts/tools/sse-load-test.js
ticket:createnode scripts/global/ticket-create.js
toolchain:readability:installnode scripts/global/install-readability-toolchain.js
validate:compatnode scripts/validate-plugin-compat.js
validate:triagenode scripts/validate-plugin-triage.js
wiki:annealnode scripts/wiki/anneal.js
wiki:auto-updatenode scripts/wiki/auto-update-pipeline.js
wiki:driftnode scripts/wiki/drift-detector.js
wiki:ingestnode scripts/wiki/ingest.js
wiki:ingest:codenode scripts/wiki/ingest-code.js
wiki:lintnode scripts/wiki/lint.js
wiki:reindexnode scripts/wiki/reindex.js
wiki:replay-evalnode scripts/wiki/replay-eval-harness.js
wiki:searchnode scripts/wiki/search.js
worktree:bootstrapbash scripts/worktree-bootstrap-node-modules.sh
worktree:startbash scripts/worktree-session-start.sh

Token telemetry reconciliation configuration

  • Run npm run routing:reconcile to generate logs/token-telemetry-reconcile.json.
  • Default thresholds: drift_pct=0.15, drift_pct_fail=0.35, min_samples=3.
  • OpenRouter aggregate usage is auto-probed when OPENROUTER_API_KEY is set.
  • Anthropic/LiteLLM aggregate usage can be enabled with ANTHROPIC_USAGE_URL and LITELLM_USAGE_URL (plus keys).
  • OpenAI-compatible providers should be logged generically unless a narrower adapter is known.
  • Codex VS Code sessions are first-class routing consumers, but do not assume they expose per-request token totals; reconcile route metadata with aggregate OpenAI usage or configured Codex OpenTelemetry exports.

Public trust surfaces

Runtime mapping

SourceRuntime target
skills/~/.copilot/skills + ~/.agents/skills
instructions/~/.copilot/instructions
hooks/~/.copilot/hooks + ~/.codex/devenv-ops/hooks
scripts/global/~/.copilot/scripts + ~/.codex/devenv-ops/scripts
.codex/~/.codex/AGENTS.md + config.toml + hooks.json + rules/
wiki/~/.copilot/wiki + ~/.codex/devenv-ops/wiki

Issue flows

Formerly DevEnv Ops. Codex name rejected due product conflict; Aegis rejected due broad name reuse.

/xteam — cross-runtime synthesis slash command

Kick off cross-team R&D synthesis across all 4 AI runtimes with a single slash command. After installing the MCP server per runtime, type /xteam <epic-N> or /xteam-create <description> in any team's chat.

See docs/howto/cross-team-rd-synthesis.md for the full workflow.

Cross-team R&D synthesis

For Epics that benefit from multi-team independent perspectives, see How-to: Run a cross-team R&D synthesis. One-line kickoff: npm run synthesis:init -- --epic <N>. Canonical protocol at instructions/cross-team-rd-synthesis.instructions.md. Tier-1 by default; HAMR R2 mailbox is the Tier-2 optimization.

Ver o original no GitHub

Bom para

Relacionados