WorkGraph
The system of record for autonomous engineering work: objective, agents, dependencies, tool calls, files, commits, risks, approvals, and proof state.
WorkGraph + Governance Core
The system of record for governed autonomous engineering work.
WorkGraph records objectives, agents, dependencies, authority, tools, files, commits, checks, approvals, risks, rollback, and proof state so coding agents become an auditable engineering workforce instead of isolated task runners.
Governance Core
The authority layer for agent identity, branch/path ownership, execution locks, policy gates, tool permissions, escalation, and human approval.
Proof Packet
The trust artifact for every governed run: plan, trace, checks, artifacts, risks, approvals, rollback path, reviewer checklist, and handoff state.
Runtime Fabric
Adapters that treat Claude Code, Codex, OpenCode, OpenClaw, Hermes, and future agents as swappable workers under one control plane.
ToolMesh
Permissioned access to GitHub, CI, Linear/Jira, Slack, Kubernetes, Terraform, observability, documents, browser, and internal systems.
Workforce Memory
Persistent doctrine, repo conventions, incident history, failed attempts, proof templates, policy decisions, and reusable execution heuristics.
Contract objects
Every autonomous run becomes structured evidence.
The WorkGraph is not chat history. It is the operational ledger for work ownership, dependencies, side effects, policy decisions, and proof-backed handoff.
ObjectiveThe business or engineering outcome, acceptance criteria, owner, constraints, and approval class.
AgentA specialist worker identity with role, runtime, permissions, authority scope, and current assignment.
TaskA unit of work with dependencies, lease state, required checks, artifacts, and handoff expectations.
Execution receiptA signed record of tool calls, commands, file changes, outputs, logs, errors, and side effects.
Governance decisionPolicy result, approval state, escalation reason, blocked action, or human override.
Proof PacketThe final evidence bundle that turns autonomous execution into a reviewable organizational artifact.
01proposed
The objective or task exists, but no agent has authority yet.
02leased
One agent owns a task/path/branch scope for a bounded time and purpose.
03blocked
Policy, conflict, missing context, failed check, or approval requirement stops progress.
04verified
Required checks, review, receipts, and risk classification are complete.
05approval_required
A human must approve PR handoff, merge, deployment, rollback, or escalation.
06superseded
A newer run or decision replaced this task, artifact, commit, or recommendation.
CI failure to verified PR handoff proves the system. It is not the system.
Each sample workload is a way to demonstrate governance, coordination, verification, proof, and approval. The product remains the governed workforce substrate.
Verified PR Handoff
A sample proof workload: diagnose, patch, verify, review, packetize, and ask a human to approve.
Verified PR Handoff
A product issue becomes a scoped WorkGraph with checks, risks, rollback, and reviewer-ready evidence.
Verified Fix Handoff
A finding is routed through policy-aware remediation, regression checks, and approval gates.
Root Cause + Patch Packet
A production signal becomes diagnosis, proposed fix, rollback path, and escalation evidence.
Risk-Checked PR
A package change is assessed, implemented, tested, and summarized with compatibility risk.
Verified Optimization
A regression is profiled, improved, benchmarked, reviewed, and handed off with rollback notes.