High-risk tools in Genesis
38 of the 1127 tools in Genesis are classified as high risk. This page profiles those tools specifically, with recommended policy actions and the attack patterns that target them.
Every operation listed below is an action PolicyLayer recommends controlling at the transport layer. Open any tool to see the full profile, risk score, and YAML policy snippet.
Tools at high risk
-
auto_startActingExecuteStart an acting appointment for a position (PUT /v1/auto/positions/{position}/acting)
-
brews_compileBrewExecuteCompile a Brew for a target runtime (POST /v1/brews/compile)
-
brews_runBrewExecuteRun a Brew (answers 503 BREWS_E_METERING_UNAVAILABLE through Genesis) (POST /v1/brews/run)
-
brews_runBrewBenchExecuteRun a Brew's benchmark (answers 503 BREWS_E_METERING_UNAVAILABLE through Genesis) (POST /v1/brews/bench/run)
-
dials_deployWorkflowExecuteSave a normalized workflow and enabled routing descriptor (POST /v1/dials/workflows/deploy)
-
fleets_launchAgentExecuteLaunch a persistent Henry agent (POST /v1/fleets/agents)
-
fleets_startInstanceExecuteStart already admitted Henry compute (POST /v1/fleets/instances/{instance_id}/start)
-
fleets_stopAgentExecuteStop a persistent Henry agent (POST /v1/fleets/agents/{agent_id}/stop)
-
fleets_stopInstanceExecuteStop admitted Henry compute and retain its data (POST /v1/fleets/instances/{instance_id}/stop)
-
flocks_executeAgentExecuteExecute a task on a canonical agent through Runner admission and a Roost node (POST /v1/flocks/agents/{agent_id}/execute)
-
foundry_aggregateRoutingEvalExecuteBuild an eval run from per-item outcomes (POST /v1/foundry/routing/evals/aggregate)
-
gears_signalRunExecuteSend a signal to a running workflow (POST /v1/gears/runs/{id}/signals)
-
gears_startChildRunExecuteStart a child workflow run under a parent run (POST /v1/gears/runs/{id}/children)
-
gears_startRunExecuteStart a workflow run (POST /v1/gears/runs)
-
gears_startTimerExecuteStart a durable timer on a run (POST /v1/gears/runs/{id}/timers)
-
henry_commandRunExecutePause, resume, cancel or send follow-up input (POST /v1/henry/runs/{run_id}/command)
-
henry_forkRunExecuteStart a new run from a past turn of this run (POST /v1/henry/runs/{run_id}/fork)
-
henry_replayRunExecuteRead every event of a Henry run and fold them into turns: each turn's inputs and the model's answer, plus pending events after the last answer. A turn's number is the `turn` to ...
-
hives_deploy_fleetExecuteDeploy a MAP fleet from a TOML manifest. (POST /v1/hives/fleets)
-
hives_invoke_deploymentExecuteInvoke a real running owned deployment and return the raw guest HTTP response (POST /v1/hives/projects/{project_id}/deployments/{deployment_id}/invoke)
-
hives_invoke_functionExecuteComplete a real function invocation and return the raw guest HTTP response (POST /v1/hives/projects/{project_id}/functions/{function_id}/invoke)
-
hives_invoke_function_asyncExecuteComplete invocation through the compatibility alias (no durable queue) (POST /v1/hives/projects/{project_id}/functions/{function_id}/invoke/async)
-
hives_marketplace_reconcileExecuteExplicitly finish an existing durable operation; this never starts a build. (POST /v1/hives/projects/{project_id}/marketplace/installed/{installation_id}/reconcile)
-
hives_stop_deploymentExecuteStop an active deployment after runtime acknowledgment (POST /v1/hives/projects/{project_id}/deployments/{deployment_id}/stop)
-
hives_trigger_cronExecuteManually trigger a cron job (POST /v1/hives/projects/{project_id}/cron/{cron_id}/trigger)
-
lanes_answerExecuteBuild a frame, route it and answer (POST /v1/lanes/context/answer)
-
minds_stopMindExecuteStop an owned Mind while retaining its volume without an automatic retry (POST /v1/minds/services/{mind_id}/stop)
-
orgs_request_formationExecutePrepare the formation packet and start a rehearsal run (or dry run) (POST /v1/orgs/entities/{entity}/formation)
-
panes_actOnPaneExecuteRun an action on one Pane (POST /v1/panes/panes/{id}/action)
-
rigs_actOnRigExecuteRun an action on one Rig (POST /v1/rigs/{id}/action)
-
runner_startJobExecuteStart a bounded Runner task with an idempotency key (POST /v1/runner/jobs)
-
shells_actOnShellExecuteRun an action on one Shell (POST /v1/shells/shells/{id}/action)
-
shells_execShellExecuteRun one command in a Shell (POST /v1/shells/shells/{id}/exec)
-
traces_pauseRunExecutePause a run (POST /v1/traces/runs/{id}/pause)
-
traces_retryRunExecuteRetry a failed run from its last failed step (POST /v1/traces/runs/{id}/retry)
-
zaps_commandExecuteRun a key-value command (POST /v1/zaps/command)
-
zaps_pipelineExecuteRun a pipeline of key-value commands (POST /v1/zaps/pipeline)
-
zaps_zapCommandExecuteRun a command in one Zap (POST /v1/zaps/zaps/{name}/command)
Attacks that target this class
High-risk tools in any server share these documented attack patterns. Each links to the full case and the defensive policy.