Execution Associates Put us on it

ProjectsisbDocs

MCP tools

Everything isb serve does is a tool: a named operation with a JSON Schema for its arguments. The same tools are served as MCP (for agents), as REST (POST /api/v1/tools/<tool>), and over the daemon's unix socket (for the isb CLI), and the web UI is built on them. One authorizer judges every call, whatever the door. This page lists all of them; the HTTP API says how to call them, and each area links to its guide.

The list a caller actually sees can be shorter: --allow-tools and --deny-tools hide tools from remote callers (Configuration), and a tool hidden that way does not exist for them. GET /api/v1/tools and MCP's tools/list return what this listener offers, with each tool's schema and annotations. Both answer 401 to a caller with no credential, unless the daemon runs with --allow-unauthenticated.

Three counts differ for that reason, and none is a fault:

  • The daemon's startup line (... (N tools)) counts every tool the listener could offer: the unix socket's full list.
  • /mcp lists the same tools to a signed-in caller (a call is still judged by role and scopes).
  • An org's /orgs/<org>/mcp lists fewer: the host, superadmin and platform tools (host_*, superadmin_*, org_nesting, org_list, org_create, server_*, user_list and the like) are on /mcp only, for anyone but the unix socket. GET /api/v1/tools marks each tool with org_endpoint (false for those), and the web UI's MCP page counts the tools an org endpoint lists from it.

The canonical argument names are the ones in each schema: an app is name and a command is argv. The app_* tools also accept app for name, and the exec tools accept command (an array, or a string that runs as sh -c) for argv; the schema and the examples show only the canonical name.

Who may call what#

Every tool takes org (default default) and works in that org; an org-bound endpoint (/orgs/<org>/mcp, /orgs/<org>/api/v1/tools/...) fills it in and refuses any other value. Then, in order:

Caller May call
the unix socket, a superadmin every tool, in every org
a platform admin every tool but the superadmin ones, in every org
an org owner or admin every tool of the org; plus managing members and tokens, reading the audit log
an org member every tool of the org, except where the table says admins
an org viewer read-only tools only, and none that hands out secret material
anyone else nothing (a token or session for another org gets "no access to org")
  • Read-only means the tool is annotated readOnlyHint and is not a secret read. secret_get, secret_resolve and app_webhook are secret reads, and so is database_get with reveal: true; viewers and read or deploy tokens are refused them, and every call is in the audit log.
  • Platform tools reach across orgs and are for platform admins only, whatever the caller's role in an org (an org owner's token is refused): org_list, org_create, org_update, org_delete, every server_* tool, server_status, registry_gc, notification_settings, template_catalog_add, template_catalog_remove, audit_verify, user_list and user_update. secret_reencrypt with all: true too.
  • Cross-org reads (overview, events, stack_list, ingress_status) are open to anyone signed in and show only the caller's orgs. audit_list and history_query filter themselves the same way.
  • Superadmin tools (host_inventory, host_policy, superadmin_token_list, superadmin_token_revoke, org_nesting) are refused to everyone else, platform admins included.
  • API token scopes narrow a token below its role: read (read-only tools), deploy (read plus stack_deploy, stack_redeploy, stack_rollback, stack_scale, app_scale, app_restart, instance_restart, app_deploy, app_rollback, build_run), admin (the whole role) and tool:GLOB. See Scopes.
  • Account tools (below) are judged by the identity endpoints' own rules rather than a role in org: a workspace token reaches none of them but whoami, a token scoped short of admin only reads them, and a token never mints a token.
  • Some tools check more themselves; the Who column says so.

In the tables, Who is the least role that may call the tool: viewer means every member of the org, member means members, admins and owners.

Stacks#

Stacks.

Tool Who Does
stack_deploy member Deploy or update a stack from compose YAML (name, compose, vars, secrets, base_dir, wait, timeout, dry_run). Returns the change per service; wait blocks until it settles; dry_run returns the changes without deploying.
stack_list anyone signed in Every stack in the caller's orgs with its services' replica, health and rollout state.
stack_status viewer One stack in detail: per service its revision, state, message, every replica (status, health, IP, in rotation, restarts, last probe output), published ports with their backends, and each domain with its URL and certificate state.
stack_config viewer The deployed compose file, and its secrets as references (store name, driver, version), never values.
stack_export viewer The compose file a stack runs from as YAML text for stack_deploy (what the web UI's stack editor shows): resolved, with file/environment secrets named as external store secrets so no value is needed. Also managed_by (apps for a project environment's stack) and the services.
stack_validate viewer A dry run of stack_deploy for an editor: {valid, errors: [{line, column, message}], changes, exists, managed_by, diff}, where diff is a unified diff from the deployed file. A bad file is an answer, not an error. Writes nothing.
stack_logs viewer Recent output of a service's replicas (slot, lines): the supervised command's journal, or an OCI image's console; last_failed_attempt as for app_logs.
stack_scale member Set a service's replicas (0 stops it without removing it).
stack_redeploy member Replace a service's replicas though nothing changed: a moved tag, changed bind-mounted files.
stack_rollback member Back to the previous deployment; a second rollback undoes the first.
stack_remove member Delete a stack's instances and ports (volumes with volumes: true), and the <stack>_<key> secrets it stored that no other stack uses.

Instances#

isb for kubectl users: look inside what runs and act on it, by app or by instance. An instance is anything isb manages in the org: an app's replica, a database, a compose service's replica, the workspace, a sandbox, a build, an org's tunnel.

Tool Who Does
instance_list viewer Every instance of the org, one row each: kind (app, database, stack, tunnel, workspace, sandbox, build), owning app or stack and service, slot, revision, status, health, in_rotation, ip, restarts, created_at, age_s, cpu_pct, mem_bytes. Filters: app, stack, service, kind, status.
instance_get viewer One instance in full: the row plus image, limits, environment variable names (never values), volumes, devices, ports, the domains its service serves and whether traffic reaches it, the last health probe, the files isb delivers into it, labels, and its recent history (history sets how many rows).
app_exec member Run argv in one of an app's replicas (default: a running one, healthy and in rotation first; replica is the slot, or instance its name): exit code, stdout, stderr. stdin (at most 1 MiB), cwd, user, env, timeout (default 60s, at most 15m; timed_out and the output so far when it runs out). Each stream keeps its last 1 MiB (truncated). The audit row keeps the argv, the names of env and the size of stdin.
instance_exec member The same in any running instance of the org by name (a workspace runs as its user, in its home).
app_logs viewer An app's replicas' recent output by app name (replica, tail default 200, since such as 10m); stack_logs underneath. While the app is not converged, last_failed_attempt carries the output of the last replica that failed to come up (instance, reason, output) after it was deleted.
app_top viewer Per replica CPU, memory, disk and network counters now, and the totals next to the app's limits (history: true adds CPU samples).
app_events viewer Events about one app, newest last (since, limit, stack_wide: true adds the stack's own).
app_restart member A rolling replace of the app's replicas with fresh instances of the same settings, in update_config's order (wait, timeout).
app_scale member The replica count (0 stops it); also saved as the app's replicas setting.
instance_restart member Replace one instance: a replica is deleted and the controller makes its replacement (wait); a sandbox restarts; the workspace has workspace_restart.
instance_file_read member One file of an instance, at most 4 MiB: utf8 or base64 (encoding). A secret read: always audited (path, never content). Not /proc, /sys, /dev or a workspace's token.
instance_file_write member Replace one file (content, encoding, mode, uid, gid, parents), at most 2 MiB. Refused: /run/isb, /run/secrets, /etc/isb, files isb delivers from org secrets, /proc, /sys, /dev. Audited with path and size.

There is no port-forward tool: run curl with app_exec, or reach an instance's address from the workspace.

Sandboxes and workspaces#

Workspaces and sandboxes.

Tool Who Does
sandbox_create member Create or reconcile one sandbox from a service spec (an object or YAML, container_name set), with expires and idle_timeout (the org's defaults otherwise: 24h and 2h) and wait_ready. Remote callers' specs are held to the remote-spec policy, and their sandboxes labelled isb.owner=<identity>. spec.egress confines the sandbox's network to a list of hosts (or none) and gives it secrets that never enter the guest; a secret the org does not hold is refused.
sandbox_list viewer Instances, filtered by labels and kind (sandbox, workspace, replica, build): creator, age, expiry, idle timeout, last activity, limits and use, and mine. Remote callers see only instances isb manages.
sandbox_exec member Run argv in a sandbox: exit code, stdout, stderr (each capped at 256 KiB, keeping the end), optional stdin text, user, cwd, env, timeout (default 10m).
sandbox_extend the sandbox's creator, or admin Push the expiry out (by, default 24h, at most 30 days from now) or change idle_timeout.
sandbox_remove member Delete a sandbox (not a stack replica, nor the workspace).
ssh_host_keys viewer An instance's SSH host public keys, for pinning, and the user isb ssh-config logs in as by default.
workspace_get viewer The workspace (or null) and the org's settings: definition, status, resources, home, live sessions, last activity, token metadata, connect (url, mcp_url), sandbox count.
workspace_list viewer Every workspace in the org.
workspace_create admin image (default isb-workspace, then dev-base, where the host has it, else images:ubuntu/24.04), name, user, cpus, memory, root_size, home_size, env, secrets, labels, token_role, setup (a first-boot script); home_bind (a host folder as the home) is for superadmins.
workspace_update admin Any of those but name, user, home_bind; resizing needs confirm: true.
workspace_start member Start it and deliver its credentials.
workspace_stop, workspace_restart member Without confirm: true, only report the live sessions it would end.
workspace_rebuild admin A fresh machine from its image (or image), same home and token; confirm.
workspace_delete admin The machine and its token, and the home unless keep_home; confirm.
workspace_token_rotate admin A new token, delivered inside; the old one stops at once.
workspace_settings member reads, admin changes sandbox_expiry, sandbox_idle; max_workspaces, home_kind, home_pool are for platform admins.
workspace_setup_run admin Run the first-boot script again, as root: now when running, else on the next start; outcome and output in the history (workspace.setup).
workspace_terminals member The web terminal's mode (herdr or shell) and its herdr sessions.
workspace_terminal_update member A herdr session: rename, or end: true.
workspace_image_build platform admin Build a workspace image from a recipe (default: isb's, as isb-workspace) in a throwaway container in isb-system; returns an id. Workspace images.
workspace_image_logs platform admin A build's state and log lines (since, wait).
workspace_image_list platform admin The images isb built, recent builds, and whether the default image is current.
workspace_image_remove platform admin Remove an image isb built (never another).
workspace_port_list member The published ports: each with its preview host, and its ingress hostname, URL and state.
workspace_port_add member Publish port; with host (a hostname, default or auto) also through the org's ingress.
workspace_port_remove member Stop publishing port.
workspace_port_open member A one-time link (60 s) to the port's preview; origin when the daemon has no --preview-domain.

Secrets#

Secrets. Values travel base64.

Tool Who Does
secret_create member Create (name, value, optional driver, labels); fails if it exists.
secret_set member A new value and version (creates it in the local store if missing); rolled lists the stacks rolling to it.
secret_get member {meta, value}. A secret read: always audited.
secret_list viewer Metadata, never values; used_by per secret, and references: the driver references stacks use.
secret_inspect viewer One secret's metadata.
secret_delete member Refused while a deployed stack uses it.
secret_refresh member Re-read from an external source (a store name, or a stack's driver reference); rolled lists the stacks rolling.
secret_reencrypt member (platform admins for all: true) Re-encrypt to the current recipients.
secret_recipients viewer The public keys values are encrypted to: the daemon's, then the break-glass ones.
secret_resolve local callers only The values of a compose file's external/age/driver secrets, for isb up.

Projects, apps and previews#

Deploy apps, Preview deployments.

Tool Who Does
project_create member A project (name, description, environments, default production).
project_list viewer Projects with each environment's stack and apps.
project_delete member A project with no apps.
environment_create, environment_list, environment_delete member (list: viewer) A project's environments; one with apps cannot be deleted. Each runs its apps as the stack <project>-<env>.
app_create member An app: an image or a git source with a builder, plus env, domains, volumes, files, ports, replicas, port, health check, resources, command, user, working directory, previews (deploy: true deploys it too). Returns the app and its webhook secret.
app_get, app_list viewer Settings, stack, service name, current deployment, webhook path, ingress_enabled; env with {secret: NAME} references.
app_update member A merge patch of settings (null clears one); takes effect at the next deploy (deploy: true). The result has a warning when the app has domains and the server runs without an ingress.
app_export viewer The app as a YAML (or format: json) document: the fields app_create takes, secrets by name only. See An app as YAML.
app_apply member Replaces the app's whole definition from a definition (YAML or JSON text, or an object): anything the document leaves out is reset to its default. To change a few fields use app_update; to edit fully, app_export, edit, dry_run, apply. Creates the app when the name is new. Name, project and environment cannot change. Refused for an existing app when it would remove domains, env vars, volumes, ports or files, or reset a port, healthcheck, resources, command, previews, user or working_dir, unless allow_removals: true; the error lists them. dry_run answers {valid, errors: [{line, column, message}], action, changes, removals, diff} and writes nothing; deploy queues a deploy after. Audited under the app's name.
app_delete member Its service leaves the stack; records, checkout, webhook secret and deploy key go; named volumes are kept.
app_deploy member Queue a deployment (wait, timeout).
app_rollback member Queue a deployment of an earlier one's image and settings, without building (deployment, wait).
app_deployments viewer The history, newest first (limit).
app_deployment_log viewer A deployment's log from a byte offset, with its record.
app_env_get viewer The environment as .env text; secret references, never values.
app_env_set member Replace it with .env text (deploy).
app_webhook member The webhook path and secret (rotate). A secret read.
app_deploy_key member A new ed25519 deploy key; returns the public half.
preview_list viewer Previews of one app (name) or of every app in the org.
preview_get viewer One preview (name, number) with its deployments.
preview_log viewer A preview deployment's log from an offset.
preview_redeploy member Fetch the head again, build, roll (wait).
preview_delete member Remove one now (wait).

Templates#

Templates.

Tool Who Does
template_list viewer The built-in catalog and added ones; filter by query words, tag, catalog.
template_get viewer A template's variables, apps, notes and, for a Dokploy or Coolify template, its translation report (compatibility).
template_deploy member Deploy into a project environment as apps (template, project, environment, name, values, dry_run, wait, timeout); without wait, the first app's deployment comes back as first_deployment.
template_instance_list viewer Deployed templates: apps, secret names, non-secret values, URLs.
template_instance_delete member Its apps (named volumes kept) and its tpl.<name>.* secrets.
template_catalog_list viewer The added catalogs.
template_catalog_add, template_catalog_remove platform admin Add (or replace) or remove a catalog: a host directory or an https URL, native, dokploy or coolify format.

Databases and backups#

Databases, backups and restores.

Tool Who Does
database_create member A database app (engine, version, database, user, publish, env, resources), deployed unless deploy: false (wait).
database_list viewer Databases with connection details, the password as a secret reference.
database_get viewer; member with reveal One database; reveal: true adds the password and URL values (a secret read).
backup_destination_create member An S3-compatible bucket; key pairs kept as org secrets; tested unless test: false. An endpoint on the daemon's own host (loopback, link-local) is for local callers and platform admins only.
backup_destination_list viewer Destinations, key pairs as secret names.
backup_destination_delete member A destination no backup uses, with the key secrets isb stored for it.
backup_destination_test member Write, HEAD and delete a small object.
backup_create, backup_update member (admin for a volume) A schedule for a database or a named volume: cron, timezone, keep (7), compression (gzip, zstd, none), enabled, missed_grace.
backup_delete member (admin for a volume) The schedule and its records; files stay in the bucket.
backup_list viewer Schedules with last and next run; with name, the backup's files in the bucket.
backup_run member (admin for a volume) Back up now (wait, timeout).
backup_runs viewer A backup's runs, or the org's restores with restores: true.
backup_run_log viewer A run's log from an offset (restore: true for a restore run).
backup_restore member Restore a database backup into target (needs confirm: true) or new. Volume backups restore with volume_restore.

Volumes#

Volumes. Every change here is for the org's admins and owners.

Tool Who Does
volume_list viewer The org's named volumes: instances using each, snapshot schedule, staged restores.
volume_get viewer One volume: instances, settings and next run, snapshots, staged restores, backups of it.
volume_snapshot_list viewer Snapshots, newest first, with kind (auto, manual, other).
volume_snapshot_create admin Snapshot now (snapshot name, wait, timeout), after the pre-snapshot hook.
volume_snapshot_delete admin Delete a snapshot.
volume_snapshot_schedule admin A merge patch: schedule (null removes it), timezone, keep, enabled, missed_grace, hook_timeout, hook_required.
volume_snapshot_runs, volume_snapshot_run_log viewer Snapshot runs and their logs.
volume_restore admin Restore staged from a snapshot, a backup (+key) or destination + key, into a new volume at /restore/<stamp> (instance, wait).
volume_restore_list viewer Staged restores.
volume_restore_discard admin Detach and delete one by stamp.

Jobs#

Scheduled jobs.

Tool Who Does
job_create member name, schedule, timezone, target ({app} or {stack, service}), mode (exec, run), command, timeout, concurrency, keep, enabled, user, cwd, env, missed_grace.
job_list, job_get viewer Jobs with their next and last run.
job_update member A merge patch; the name is fixed.
job_delete member The job and its run records (refused while it runs).
job_run member Run now (wait, timeout).
job_runs, job_run_log viewer The history; one run's output from an offset.

Builds and the registry#

Builds and the local registry.

Tool Who Does
build_run member Build a host directory (app, context, subdir, builder, dockerfile, target, args, tag, untrusted, timeout) into the org's registry repository; returns an id. A remote caller's context must be under a --bind-root.
build_logs viewer A build's state and log lines from since, waiting up to 30 s (wait) for more; image and digest when it succeeded.
build_list viewer The org's recent builds.
registry_list viewer The org's images: apps, tags, digests, push times.
registry_gc platform admin Keep the newest keep tags per app (10) and whatever deployed stacks run or would roll back to; dry_run.

Notifications and metrics#

Notifications, Uptime monitoring, Metrics.

Tool Who Does
notification_channel_create member name, provider (webhook, Slack, Discord, Telegram, email; secrets named, never values), rules, enabled.
notification_channel_list, notification_channel_get viewer Channels with their last delivery.
notification_channel_update member Replace provider, rules or enabled; fields left out are kept.
notification_channel_delete member The channel and its delivery log (its secrets stay).
notification_test member Send a test message now; returns the delivery.
notification_deliveries viewer A channel's last 50 deliveries, newest first.
notification_settings platform admin allow_private_targets, server-wide.
monitor_create, monitor_update member An uptime monitor: name, type (http, tcp, app), its target, expected_status, keywords, headers (values or secret names), interval, timeout, thresholds, cert_expiry_days (Uptime monitoring).
monitor_list, monitor_get viewer Monitors with status, last check, uptime (24 h, 7 d, 30 d), latency p50/p95, bars; the org's incidents and settings.
monitor_checks viewer A monitor's history over range: buckets, uptime, raw checks.
monitor_delete, monitor_pause, monitor_resume member Remove (with its history), stop or start checking.
monitor_settings member auto_monitors (apps with a domain get their own monitor), exclude_apps.
metrics_query viewer Metrics history: metric, app or stack/service or instance, range or from/to, step, aggregate.

Orgs, the ingress and the dashboard#

Orgs, Domains and ingress.

Tool Who Does
org_get viewer Limits and per-instance defaults, bridge and subnet, egress exceptions, bind roots, service-name domain, counts, and placement (kind, server, isolation).
org_list platform admin Every org, as org_get shows one, with the server it runs on.
org_create platform admin org, cpus, memory, disk, instances, default_cpus, default_memory, egress, placement ("local", {"server": NAME}, {"vm": {cpus, memory, disk}}), wait. Bind roots are set on the host only.
org_update platform admin Limits, defaults or egress (which replaces the list; [] clears it); a different placement is refused.
org_delete platform admin Refused while stacks are deployed; force deletes remaining sandboxes; delete_vm deletes a dedicated VM.
ingress_status anyone signed in Listeners, CA, the Caddy process, every routed domain (URL, certificate state, upstreams), conflicts and refusals, each tunnel org's cloudflared. The caller's orgs only.
overview anyone signed in Everything a dashboard shows in one call: host CPU and memory with history, every stack in detail, sandboxes with CPU and memory, the latest event number.
events anyone signed in The event feed after a since cursor (limit), waiting up to 30 s (wait) for one.
server_status platform admin isb's and incus' versions, and the balancer's routes with live counters.

Audit and history#

The audit log, The history.

Tool Who Does
audit_list org owner or admin (their org); platform admin (everything) Entries filtered by actor, action, target (globs), outcome, surface, user_id, token_id, since/until, platform; paged with before, tailed with after.
audit_verify platform admin Walk the audit log's and the history's hash chains.
history_query viewer (their orgs); host-level rows: platform admin Controller events, incus lifecycle events, audit rows (owners and admins) and markers, merged; filter by object (exact), kind, source, actor, since/until, platform; correlate links incus changes to their likely audit row.

Servers#

Servers and dedicated VMs. All for platform admins.

Tool Does
server_add Bootstrap a box over SSH (name, ssh, ssh_port, key (a path, local CLI only) or ssh_key (the key itself), address, agent_port, allow_from, isb_binary, version, self_binary, public_ingress); wait: false answers at once.
server_list Servers with health, orgs and version (build, protocol, skew against this control plane, upgradable, last upgrade); servers being added (provisions); dedicated_vm (whether this host can run dedicated VMs); suggested_allow_from.
server_show One server: address, how it was added, certificate fingerprint and expiry, health, orgs.
server_remove Forget a server (refused while it holds orgs; a dedicated VM is deleted with it).
server_rotate_cert Issue its agent a new certificate.
server_upgrade Replace a server's agent (name, or all: true) with this control plane's build, a release (version) or a binary on this host (isb_binary, local CLI only); waits for the new build, and the box rolls back if it does not answer.
server_provision_get Follow a server (or a dedicated VM, vm-<org>) being added: steps, log, state, error.

A call for an org placed on a server is judged on the control plane, then again by the server's agent. events, audit_list, audit_verify, registry_gc, notification_settings, server_status, template_list, template_get, the template_catalog_* tools, the server_* tools and the account tools always run on the control plane.

Accounts#

Users, roles and superadmins. The identity endpoints' account pages as tools, with the same rules (Identity API): a workspace token reaches none of them but whoami; a token scoped short of admin only reads them; nobody outside an org learns about it (not_found); only an owner (or a platform admin) touches an owner or makes one.

Tool Who Does
whoami anyone signed in, a workspace included The caller: user, platform admin flag, orgs and roles, how it signed in (auth), the orgs it can open, and superadmin.
member_list viewer Each member's user (id, email, name), role and last activity.
member_update admin user_id or email, role.
member_remove admin; anyone for themselves user_id or email. Their account stays.
invitation_list admin Pending invitations.
invitation_create admin email, role (default member, at most your own). Returns the invitation, its token (shown once) and the link when --public-url is set. isb sends no mail.
invitation_revoke admin id.
agent_identity_list viewer The org's tailnet and Access agent identities (kind, subject, role, note) and available: the server's tailnet listen addresses, whether Access guards a listener, its public URL, and reach (who gets in with no mapping: counts, and names for owners and admins only).
agent_identity_set admin kind (tailnet, access), subject (a tailnet login or tag:name; an Access service token client id or the email of someone who is not an isb user), role (viewer, member, admin: never owner, at most your own), note. An existing subject changes role.
agent_identity_remove admin id.
token_list anyone signed in Your tokens' metadata (only org's when given; an org token sees its org's). all: true: every token in org, with its holder (admins).
token_create a session, or an Access or tailnet identity name, expires (90d; none: never), scopes; confined to org. Returns {token, info}, the token shown once. A token cannot mint tokens (an API, workspace or superadmin token is refused), so revoking a leaked token always ends it.
token_revoke its holder; the org's admins id.
ssh_key_list, ssh_key_add, ssh_key_remove anyone with an account The caller's SSH keys for isb ssh-proxy: public_key and name; id.
session_list, session_revoke anyone with an account The caller's browser sessions; end one by id.
user_list platform admin Every user with their orgs and last activity.
user_update platform admin user_id or email; disabled, platform_admin. Not yourself, and never the last enabled platform admin.

Signing in (passwords, passkeys, providers), sign-up, password resets, accepting an invitation and changing a way in stay in the browser, and superadmin tokens are minted on the host only.

Superadmins#

Users, roles and superadmins.

Tool Does
host_inventory Every incus project and instance on the host, isb's or not: project, org, type, status, addresses, isb's stack and owner labels.
host_policy How the daemon serves: listen addresses, Access, the remote tool policy, what remote specs may ask for, and each superadmin source with its allow list and token count.
superadmin_token_list Superadmin tokens' metadata, never the token.
superadmin_token_revoke Revoke one by id. Minting is isb token create NAME --superadmin, on the host only.
org_nesting Read (org) or set (allow_nesting) whether the org's workspace may run Docker with security.nesting (The Docker exception). Turning it off is refused while the workspace runs with nesting.