A documentation surface for the Plan → Run → Verify → Review → Approve flow for enterprise AI coding agents: engineering templates, repo authority, run evidence, delivery graph, governed tools, approvals, rollback, and execution targets. The live product console stays on cloud.agentfoundry.in.
source.readback — workflow contracts, evidence rails, and operator patterns loaded for documentation preview
okobjective.bound — issue-to-handoff documentation slice loaded for review
okruntime.attach — Execution host, orchestration, verification, planning, and template lanes selected under one contract
oktoolmesh.policy — shell, browser, governed tool access, evaluation, and sandbox actions require scoped receipts
warnreview.review — typecheck, lint, build, API tests, capability readbacks, rollback refs staged for reviewer
holdapproval.wait — external sensitive action and deploy remain blocked until human approval
blockedAgentFoundry docs track the readiness path across category fit, autonomous coding-agent capability, enterprise controls, traction evidence, and differentiated strategy.
Scope, policy state, tool actions, artifacts, owner notes, cost, rollback relevance, verification, and run evidence stay visible before PR or release handoff.
Issue or failed check → plan → patch → checks → run evidence → owner approval → PR-ready handoff.
Each section documents how engineering teams plan runs, scope repo authority, govern tools and actions, inspect run evidence, manage approvals, and improve engineering templates without widening into non-engineering automation.
12:10:08source.readbackworkflow contracts, evidence rails, and operator patterns loaded for documentation preview
ok12:10:31objective.boundissue-to-handoff documentation slice loaded for review
ok12:11:04runtime.attachExecution host, orchestration, verification, planning, and template lanes selected under one contract
ok12:11:36toolmesh.policyshell, browser, governed tool access, evaluation, and sandbox actions require scoped receipts
warn12:12:02review.reviewtypecheck, lint, build, API tests, capability readbacks, rollback refs staged for reviewer
hold12:12:48approval.waitexternal sensitive action and deploy remain blocked until human approval
blockedissue + repo + owner + acceptance criteria
livetask DAG + branch/path leases
leasedgoverned execution lanes
readygoverned tools · shell · browser · evaluation scopes
reviewtrace · checks · risks · rollback
compilingapprove · retry · narrow · stop
waitingplan/run/verify/review/approvegoverned runtime lane · receipt-backed execution
plan/run/verify/review/approveindependent verification lane · checks and replay
plan/run/verify/review/approverollback-aware recovery lane · pause and resume
plan/run/verify/review/approvespecialized implementation lane · bounded autonomy
plan/run/verify/review/approveparallel background lane · governed fan-out
plan/run/verify/review/approvereview-focused lane · evidence consolidation
plan/run/verify/review/approveisolated trial lane · policy-gated execution
plan/run/verify/review/approvecustomer-scoped lane · explicit approvals required
WorkGraph event cursor, runtime receipts, tool calls, actor identities, and capability refs.
6/6 required lanes presenttypecheck, lint, build, diff hygiene, and capability-contract assertions.
validation gate requiredpatches, logs, screenshots, generated files, capability readbacks, and proof report refs.
artifact rail visiblebranch race, path ownership, external side effects, and deployment blast radius.
human review requiredbaseline ref, revert command, supersession marker, replay cursor, recovery owner.
rollback path pinnedProof Report must be accepted before external or sensitive action
approval_requiredhuman_reviewerNo external send, production mutation, or customer-visible change from the console
blockedowner onlySpecialized execution lanes require explicit policy and spend scope
reviewgovernance_coreRelease handoff needs health checks and rollback review
holdrelease_handoffDenied until explicit command, target, and rollback are present
blockedtoolmeshorigin/mainbaselinelast fetched remote before autonomous work
anchorworkgraph://verified-prstate checkpointobjective, leases, runtime choices, policy decisions
readyreview://report/sampleevidence bundlechecks, logs, trace, artifacts, reviewer checklist
compilingartifact://patchreversible diffsupersede or revert before sensitive action
reviewevent://replayreplay cursorresume from last verified event if runtime fails
waitingThese screens illustrate run flow, approval logic, tool governance, and evidence structure without exposing internal endpoints or customer activity.
Use the product console for approved operator work. The public docs stay focused on workflow contracts, governance, and review surfaces.