All servers / Other / Call Telemetry MCP server
Call Telemetry MCP server
Expanded Model Context Protocol (MCP) server for Call Telemetry lab, fleet, and appliance management across Proxmox VE (ESXi is retired).
Tested by ConnectATDScore 89 out of 100. Tested 2026-11-18 on package version 0.8.184. Package: @calltelemetry/ct-lab-mcp · source
Score breakdown
| Runs | 18 / 20 | |
|---|---|---|
| Tool descriptions | 19 / 20 | |
| Package safety | 20 / 20 | |
| Maintenance | 20 / 20 | |
| Adoption | 12 / 20 |
What we saw
- Starts and answers the MCP handshake: yes (13.9 seconds including download)
- Tools offered: 95
- Last published: 2026-10-28
- Weekly downloads (npm): 1,965
- License: MIT · Direct dependencies: 12
Want this server set up for your team?
ConnectATD builds AI automation and wires up MCP servers like this one. Tell us what you are trying to do.
Talk to ConnectATDTools and how they are described
Agents choose tools from these descriptions. Under each name you can see how many of its parameters are explained.
lab_preflight2 of 2 parameters describedValidate canonical inventory, exact VM identity, and software release-line qualification before any guest command. This does not release software to the lab; it qualifies a fixture against a cut release.
lab_mission7 of 9 parameters describedPlan or run one typed canonical lab mission while the ct-meta authority holds the asset and coordinator locks across every phase. The manifest selects only a registered executor; arbitrary commands and arguments are not accepted.
lab_dev11 of 12 parameters describedPlan or operate one exact leased remote development workspace through the canonical lab lifecycle. Supports doctor, plan, prepare, start, status, sync, stop, and cleanup for the ct-local-dev-v1 profile; prepare only primes a bound cache and does not claim runtime readiness. Arbitrary commands, guest paths, and legacy/mock fallbacks are not accepted.
uat_access10 of 10 parameters describedPlan or establish governed web access for one exact runtime UAT appliance. Dry-run is default; apply requires the exact asset confirmation and uses only the canonical UUID-bound credential/lease workflow. No raw host, command, password, or global-credential fallback is accepted.
vm_repair2 of 10 parameters describedPlan or recover the VM-specific credential for one exact incomplete provision. Requires the original pending operation, active lease and issue, confirmed live SMBIOS UUID, and separate repair operation. Canonical local/shared locks and receipts own execution. Explicit restartConsole reconciliation can gracefully restart only this disposable VM to observe its login prompt; no force stop. Does not complete provisioning, start application services, accept guest commands, or use a generic credential fallback.
appliance_update7 of 7 parameters describedPlan or apply a release update of one exact claimed lab appliance through canonical lab-inventory vm update (the official installer over the inventory-owned pinned transport; the canonical CLI streams the update output). Dry-run is default; apply requires confirmAssetId exactly equal to assetId. Requires an active lease, a calltelemetry-lab-claim/v1 claim file authorizing 'appliance-update', an exact release version, and CT_RELEASE_REPO on the server. The only supported way to update a claimed guest: the in-guest appliance_exec flow is retired. Accepts no shell command, host, credential, or bundle pinning option.
appliance_candidate_update7 of 7 parameters describedPlan or apply a lab-only appliance candidate manifest through canonical lab-inventory vm update-candidate. Dry-run is default; apply requires the exact asset ID and manifest SHA-256 confirmation. No release publishing or local guest fallback is performed.
appliance_candidate_retire5 of 13 parameters describedPlan or retire only an exact failed/unknown disposable candidate through canonical lab-inventory. Requires matching failure, quarantine, provision and ownership evidence. Dry-run by default; apply confirms asset, VMID and UUID. Preserves failure records; never renews or reuses an expired guest.
fleet_versions3 of 3 parameters describedQualify canonical VM assets against their cut software release line. Resolves Linear claim state before any guest probe; closed claims are blocked. Requires at least one asset ID or Linear issue selector. Delegates to the canonical lab-inventory CLI.
fleet_candidates6 of 6 parameters describedRank reusable VM candidates and hypervisor host capacity against memory, vCPU, and disk requirements. Returns eligible VMs and hosts with blocking reasons for ineligible assets. Delegates to the canonical lab-inventory CLI.
fleet_group2 of 2 parameters describedResolve a canonical fleet group membership (read-only). Returns the exact member assets, host bindings, and group metadata without mutating inventory. Delegates to the canonical lab-inventory CLI.
host_maintenance_shutdown7 of 10 parameters describedPlan or apply a guarded whole-host maintenance shutdown for one exact canonical standalone Proxmox or ESXi host. Dry-run is default. Apply requires exact host confirmation plus the current plan digest, gracefully drains guests, refuses active claims or incomplete discovery, enters ESXi maintenance mode, and never offers force power-off.
cube_listno parametersList inventory CUBE assets. Exact asset ids only. Power stays vm_power. Claims stay vm_claim.
cube_status2 of 2 parameters describedInventory-bound CUBE IOS-XE status (version, interfaces, HTTP, XCC, SIP-UA). Exact assetId only.
cube_exec3 of 3 parameters describedShow-command allowlist against an inventory CUBE. Rejects reload and write erase.
cube_config6 of 6 parameters describedApply IOS config lines to an inventory CUBE. Dry-run default. Writes need apply, confirmAssetId, and snapshot receipt.
cube_dialpeer_show2 of 2 parameters describedShow existing CUBE dial-peers including 9... peers.
cube_dialpeer_apply6 of 6 parameters describedADDITIVE dial-peer apply. Must not no dial-peer existing 9... peers. Dry-run default.
cube_xcc_enable6 of 8 parameters describedAdditive XCC/WSAPI enable on an inventory CUBE. ATT-role assets are rejected. Dry-run default.
cube_xcc_status2 of 2 parameters describedRead CUBE WSAPI/XCC registrations for an exact inventory asset.
cube_ios_snapshot3 of 3 parameters describedCapture a running-config snapshot receipt required before CUBE writes. mock:true is not evidence.
cube_claim6 of 6 parameters describedClaim an interchangeable cube-emulator instance for a Linear ticket and optionally bind an appliance XCC callback URL.
credential_status1 of 1 parameter describedSelf-check: reports the resolved credential tier, which hosts/executors are real vs mock, and live reachability per hypervisor host. Call this before trusting other tool results. See QA-395.
list_hosts5 of 5 parameters describedQuery live free RAM/CPU/disk capacity, health, and deprecated resource-derived 8GB slot estimates across Proxmox VE (proxmox-lab, proxmox-mini, proxmox-ibm64) (VMware ESXi is retired and only available when ALLOW_ESXI_SLOTS=true).
list_inventory10 of 10 parameters describedQuery unified fleet inventory across Proxmox VE and VMware ESXi VMs with status, IP, VLAN, protected flag, release version, and Linear leases.
find_slot20 of 20 parameters describedDEPRECATED (slot retirement ADR (REL-1746); use list_hosts): ranks candidate hosts by live free RAM then free disk headroom and reports reclaimable expired claims and disposable VMs if the fleet is full. Slot counts are deprecated estimates. Can auto-claim when autoClaim=true, or auto-purge and claim when purgeIfFull=true.
claim_slot33 of 33 parameters describedDEPRECATED (slot retirement ADR (REL-1746); use create_appliance_claim): claim capacity on a specified or capacity-ranked hypervisor host with structured labeling, workload types, TTL, and Linear issue tracking. Capacity is admitted from live free RAM/CPU/disk, not slots. Supports auto-reclamation via purgeIfFull=true.
release_slot17 of 17 parameters describedDEPRECATED (slot retirement ADR (REL-1746); use delete_appliance_claim): release an active claim by claimId, tag, requester, or Linear issue, with optional automatic destruction of attached test VMs.
vm_renew10 of 10 parameters describedRenew one exact canonical VM lease under the canonical local and cross-host locks; running guests are allowed only through the canonical metadata-renewal proof path.
vm_transfer10 of 10 parameters describedTransfer one exact canonical VM lease to another active Linear issue; expired-running recovery is a narrow canonical proof path, never a force override.
purge_slots14 of 14 parameters describedRETIRED (slot retirement ADR (REL-1746)): returns an explicit retired error and takes no action. The operator reaper deletes expired claims and unclaimed guests; use list_expired_vms / cleanup_expired_vms.
purge_and_redeploy11 of 11 parameters describedOne-shot retirement of an old or expired disposable test VM and replacement with freshly claimed slot capacity and deployment. Verifies non-bypassable guardrails before destroying old VM.
prune_untracked_vms9 of 9 parameters describedSafely prune stopped disposable test VMs lingering in inventory without active slot leases, freeing hypervisor datastore and memory while strictly preserving protected fixtures (CUCM, CUBE, templates).
list_expired_vms3 of 3 parameters describedClassify live lab VMs as legal, expiring, expired, dead leftover, or protected. Does not delete. Terminal Linear-ticket VMs are marked for cleanup; protected CUCM/CUBE/OpenClaw/templates are never candidates.
cleanup_expired_vms10 of 10 parameters describedClassify cleanup candidates by default; dry-run may omit confirmAssets for the full plan or pass an exact scope. Mutation requires a non-empty confirmAssets list and deletes only selected delete-ready disposable VMs, including eligible Done/cancelled Linear-ticket guests. Running terminal-ticket guests receive graceful shutdown first; non-terminal running guests stay report-only. Protected fixtures are never deleted.
provision_vm64 of 64 parameters describedProvision one disposable Call Telemetry appliance on a host with verified free RAM and storage, or explicitly select remote-dev-v1 with alma97-cloud for an OS/container development guest. Development provisioning requires a reviewed offline bootstrap artifact, not an appliance release; exact interrupted development recovery after vm_repair uses this tool with the original identity and confirmations, never auto_provision_vm. Verify application readiness separately through lab_dev. Hold only for the live claim; never steal In Progress guests. For SHA-channel 0.8.8-stable use version+webImage+baseVersion or proxmox_quick_provision. Guest disks are thin-only: storage must be lvmthin (e.g. local-lvm); thick or undeterminable storage is refused with THICK_STORAGE_REFUSED before any change.
proxmox_quick_provision9 of 23 parameters describedPlan/apply fast AlmaLinux cloud provisioning across the homogeneous Proxmox fleet. Pass one size object; hostAssetId defaults to auto, which filters live memory/vCPU/disk capacity and ranks eligible hosts by free RAM then disk headroom. Apply requires the reviewed explicit hostAssetId+vmid+operationId. Protected fixtures and VMID 182 are never candidates. Guest disks are thin-only: storage must be lvmthin (e.g. local-lvm); thick or undeterminable storage is refused with THICK_STORAGE_REFUSED before any change.
proxmox_import_qcow19 of 25 parameters describedPlan or apply an appliance QCOW2 import as a disposable Proxmox lab guest through canonical lab-inventory vm provision --source published-qcow (leased, journaled, credential-recoverable). The guest boots awaiting credential enrollment: continue with vm_repair, then finalize, then uat_access. osVersion (the AlmaLinux 9 minor inside the image, e.g. 9.8) is required. imageUrl is exactly one of two identity URL shapes (no query string, no presigned URL): a published image on the download Worker host (a release candidate promoted to the public bucket keeps its -rc<N> suffix), https://download.calltelemetry.com/ct-<version>[-rc<N>].qcow2, or a private CI candidate for an unpublished RC (stored in a separate private bucket, never the public release bucket), https://<32-hex-account-id>.r2.cloudflarestorage.com/ct-ci-candidates/candidates/<version>/<run_id>/ct-<version>.qcow2. Candidates are private and digest-bound; the public publish gate is unchanged. The lab operator supplies any download credential to the canonical CLI through the gateway's own environment (CT_IMPORT_QCOW_DOWNLOAD_URL; the download host answers 403 to unsigned requests, so a published image is refused unless that URL is supplied or the gateway environment carries all of LAB_RELEASES_R2_ACCOUNT_ID, LAB_RELEASES_R2_RO_ACCESS_KEY_ID and LAB_RELEASES_R2_RO_SECRET_ACCESS_KEY (the read-only release key), from which the canonical CLI mints a short-lived presigned R2 URL itself), never through this tool. Dry-run is default; apply requires the reviewed asset, vm prefix, VMID, and image SHA-256 confirmations. VMID 182 is reserved. The canonical CLI owns digest verification, import, evidence capture, and teardown. The guest disk is thin-only: storage must be lvmthin (default local-lvm); thick or undeterminable storage is refused with THICK_STORAGE_REFUSED before any change.
claimed_import_qcow22 of 28 parameters describedImport a published or private-candidate appliance QCOW2 as a Proxmox lab guest AND claim it, in one sanctioned step. The strict appliance operator destroys every guest no ApplianceClaim names within ~5 minutes, so this tool creates a profile=bare ApplianceClaim (no seed, no license, no bootstrap; names node + VMID + VM name; finite TTL) FIRST, then runs the canonical lab-inventory vm provision --source published-qcow apply (digest-bound, leased, journaled, receipted; the guest then awaits vm_repair, finalize and uat_access), and retains finite claim protection if the import fails, throws or has an uncertain outcome. Reconcile the exact canonical operation before explicit governed teardown. Dry-run is the default: it runs the canonical import plan and previews the claim. Requires vmid, vmPrefix and node (minis|proxmox|ibm64, matching hostAssetId). A published image needs a download credential in the gateway environment, either CT_IMPORT_QCOW_DOWNLOAD_URL (a signed fetch URL) or all of LAB_RELEASES_R2_ACCOUNT_ID, LAB_RELEASES_R2_RO_ACCESS_KEY_ID and LAB_RELEASES_R2_RO_SECRET_ACCESS_KEY (the read-only release key); otherwise it is refused before the claim is created. Renew with renew_appliance_claim; release with delete_appliance_claim. The guest disk is thin-only: thick or undeterminable storage is refused with THICK_STORAGE_REFUSED before the claim is created.
pin_pool_slots6 of 7 parameters describedRETIRED (slot retirement ADR (REL-1746)): returns an explicit retired error and takes no action. There are no pool slots; pass webImage to proxmox_quick_provision to pin the web image at provision time.
auto_provision_vm66 of 66 parameters describedAuto-select the live resource-qualified Proxmox host with the most free RAM (then disk headroom), claim it, clone/boot the appliance, pin the requested release, and return verified endpoints. A test-only placementSeed makes selection reproducible. Hold until the live claim expires; cleanup returns capacity only after verified destruction. Guest disks are thin-only: storage must be lvmthin (e.g. local-lvm); thick or undeterminable storage is refused with THICK_STORAGE_REFUSED before any change.
vm_action17 of 22 parameters describedExecute VM power lifecycle operations (start, stop, reboot, shutdown, reset, status, clone, destroy) across Proxmox and ESXi with non-bypassable safety guardrails protecting core infrastructure (CUCM, CUBE, templates).
appliance_exec11 of 11 parameters describedRETIRED: the in-guest shared-login appliance CLI flow answers CANONICAL_PHYSICAL_SLOT_AUTHORITY_REQUIRED. Use appliance_update to update a claimed guest, uat_access for governed web access, and lab-inventory vm ssh (from a lab-LAN host or the ct-uat-proxmox runner) for operator shell access.
factory_reset_appliance8 of 8 parameters describedOrchestrate a complete factory reset or soft reset on an appliance VM: wipe data, switch/pin target release version, run migrations & bootstrap, apply 30-day Demo license, and assert system health.
list_storage3 of 3 parameters describedQuery storage pools, datastores, capacity, and free space across exact canonical Proxmox VE and VMware ESXi hosts. Proxmox pool content capabilities are included. Detailed volume contents require includeContent=true and remain unavailable in canonical mode, which fails closed instead of using legacy credentials.
list_snapshots6 of 6 parameters describedQuery VM snapshots, creation timestamps, descriptions, and memory states across all VMs on Proxmox VE and VMware ESXi. For a claimed bare guest pass assetId, linearIssue, leaseId and operatorClaimUid instead: this routes to canonical lab-inventory vm snapshots (read-only, claim-scoped).
clean_storage5 of 5 parameters describedDelete unreferenced disk volumes, stale ISOs, or VM snapshots with non-bypassable safety guardrails protecting core infrastructure (CUCM, CUBE, templates).
list_slots13 of 13 parameters describedDEPRECATED (slot retirement ADR (REL-1746); use list_appliance_claims and list_hosts): query claims, TTLs, expiration timestamps, structured metadata (labels, workload types, versions, branches, tags), and cluster capacity breakdown.
get_vm_storage3 of 3 parameters describedQuery detailed storage configuration for a specific VM, including hypervisor virtual disks, CD-ROM drives, mounted ISOs, and live guest OS filesystem mount points (/boot, /, /data) via QEMU agent.
mount_iso5 of 5 parameters describedMount an ISO image from hypervisor storage (e.g. 'synology-cisco-isos:iso/...' or 'local:iso/...') into a target VM CD-ROM / optical drive.
unmount_iso4 of 4 parameters describedUnmount / eject an ISO image from a target VM CD-ROM / optical drive.
create_snapshot12 of 16 parameters describedCreate a new snapshot of a target VM with optional RAM state and description. For a claimed bare guest pass assetId, linearIssue, leaseId, operatorClaimUid, operationId and snapshotName instead: this routes to canonical lab-inventory vm snapshot (disk-only, plan by default; apply needs the confirm* fields from the plan).
rollback_snapshot10 of 14 parameters describedRevert / rollback a target VM to a specified snapshot with safety validation protecting core infrastructure. For a claimed bare guest pass assetId, linearIssue, leaseId, operatorClaimUid, operationId and snapshotName instead: this routes to canonical lab-inventory vm rollback (latest snapshot made by this tooling for the claim only; plan by default).
update_appliance_cli7 of 7 parameters describedUpdate or hot-patch the remote appliance operator script (/home/calltelemetry/cli.sh) on a running appliance VM over SSH (port 2222) for rapid UAT fixes. Includes base64 delivery, syntax validation (bash -n), and status assertion.
get_appliance_diagnostics9 of 9 parameters describedInspect deep multi-subsystem telemetry across HTTP /api/health, Prometheus /metrics (BEAM memory, ETS tables, process count, DB pool), Oban job queues & fetchable depth, active alarms, and SQL connection health on a target Call Telemetry appliance.
manage_vm_network10 of 10 parameters describedInspect and manage virtual network adapters (net0, net1), MAC addresses, bridges (vmbr0, vmbr1), VLAN tags, rate limits, and link state (connect/disconnect virtual cable for partition testing).
get_guest_agent_status3 of 3 parameters describedQuery QEMU Guest Agent / VMware Tools health, agent ping latency, guest operating system and kernel version, hostname, and guest-assigned IP network interfaces.
list_cubes1 of 1 parameter describedLists all lab CUBE routers (Cat8000v), their IP addresses, power state, and current lease claim / holder Linear issue.
claim_cube8 of 8 parameters describedAcquires an exclusive lease lock on a CUBE router fixture for a test suite with automatic TTL expiration.
release_cube3 of 3 parameters describedReleases an exclusive CUBE router lease lock, returning it to the available pool.
provision_cube28 of 28 parameters describedProvisions a new Cisco Cat8000v CUBE router VM, activates advanced licensing (network-advantage with reboot), and applies clean XCC/WSAPI config.
provision_cucm_trunk17 of 17 parameters describedProvisions or synchronizes a Cisco CUCM SIP Trunk and Route Pattern pointing to a target CUBE router or media fixture via AXL SOAP with immediate reset activation.
manage_mcp_server3 of 3 parameters describedManages the MCP server runtime lifecycle: inspects status, checks for npm/git updates, triggers hot credential/cache reloads, sends tool list refresh notifications, or signals a process upgrade/restart.
get_sip_trunk_status3 of 3 parameters describedQueries real-time Cisco CUCM SIP Trunk registration status, destination reachability, and SIP OPTIONS Ping In-Service (INS) / Out-of-Service (OOS) health via RISDB / RISPort70.
setup_cucm_emulator7 of 7 parameters describedManages CUCM Emulator suite: inspects status, seeds standard lab fixtures (phones, lines, CSS, partitions, SIP trunks, JTAPI users), resets store, or triggers simulated call legs.
apply_manifest2 of 2 parameters describedApplies a declarative Kubernetes-style ApplianceSlot manifest (YAML or JSON) to automatically claim an 8GB slot lease and reconcile VM resources on Proxmox VE.
get_manifest1 of 1 parameter describedRetrieves the live status and Kubernetes-style conditions of a declarative ApplianceSlot manifest.
delete_manifest1 of 1 parameter describedDeletes/releases a declarative ApplianceSlot manifest and reclaims its slot lease on Proxmox VE.
renew_manifest2 of 2 parameters describedRenews/extends the lease duration (TTL heartbeat) of an actively applied declarative ApplianceSlot manifest.
list_manifestsno parametersLists all actively applied declarative ApplianceSlot manifests and their current phases.
apply_license19 of 19 parameters describedApplies or updates a Call Telemetry license token (Community, Demo, Essentials, Advanced, Premium) on a target appliance VM via Phoenix REST API.
renew_license19 of 19 parameters describedRenews an existing Call Telemetry license on a target appliance VM by minting and applying a fresh signed token with specified duration.
host_migrate_vm12 of 12 parameters describedPlan or execute guarded cold VM migration between standalone Proxmox hosts using native qm remote-migrate. Cold migration only: source VM must be stopped due to CPU heterogeneity.
replicate_cube_template5 of 9 parameters describedPlan or execute the claim-first CUBE template replica action for the exact protected template. The canonical path creates finite bare claims before provider writes, full-clones a stopped carrier, transfers with delete=0 and vmbr1:vmbr0, verifies the ISO-free stopped target, and cleans only the exact carrier after readback.
sandbox_clone20 of 20 parameters describedSpawn an ephemeral agent sandbox VM from gold master template in sub-15s (linked clone). Supports AlmaLinux 9.7 (authoritative CallTelemetry master) and Ubuntu 24.04.
sandbox_reset11 of 11 parameters describedRollback an ephemeral agent sandbox VM to its baseline clean-state snapshot in sub-15s.
ferrum_fleet_metrics2 of 2 parameters describedQuery fleet-wide hypervisor telemetry (free RAM, CPU and disk capacity are authoritative; legacy slot floor/max are kept as deprecated numbers) and Prometheus capacity alert readiness across Proxmox standalone nodes.
list_appliance_claims3 of 3 parameters describedLists active Kubernetes ApplianceClaim resources (dynamic & fixtures) in ct-lab with phase, IP, VMID, train, and TTL.
get_appliance_claim2 of 2 parameters describedRetrieves full details of a specific ApplianceClaim resource including status, containerOverrides, and conditions.
create_appliance_claim20 of 20 parameters describedDeclaratively creates an ApplianceClaim (claim first: the strict operator destroys any guest no claim names). Name the guest with vmid+node or vmName. profile=demo bootstraps it (license, demo seed, CUBEs); profile=bare only claims and tracks it (no seed, no license, no bootstrap) for clean benchmarks and imported images. Finite ttlHours, renewable with renew_appliance_claim; delete_appliance_claim destroys the guest.
delete_appliance_claim3 of 3 parameters describedRequests deletion of the original guarded Proxmox ApplianceClaim with expectedUid as an API precondition. The operator retains custody until independent provider, volume and claim absence are proved; recreated claims and claims with no custody record are refused.
pause_appliance_claim4 of 4 parameters describedPause or unpause an ApplianceClaim: spec.paused=true suspends TTL expiry and operator lifecycle actions for a maintenance window (deletion still outranks pause; pause never purges; ADR 0755).
renew_appliance_claim5 of 5 parameters describedExtend an ApplianceClaim's TTL so it expires extendHours from now (never shortens; fixtures and ttl=0 claims are refused). The strict operator destroys the claim's guest when the TTL elapses, so renew a long-running test before it does. When expectedUid is supplied, the renewal is guarded by exact UID and resourceVersion tests before patching.
sideload_appliance_runner6 of 6 parameters describedExecutes isolated sideloaded runner actions (execute, container_override, systemd_reset, diagnostic_bundle) over SSH without mutating in-VM scripts.
get_appliance_id3 of 3 parameters describedDiscovers host hardware UUID over SSH and computes the canonical CT-APP-XXXX-XXXX-XXXX appliance identifier.
sideload_license8 of 16 parameters describedSideloads a standardized calltelemetry-license-v2 envelope directly to /etc/calltelemetry/calltelemetry.lic with chmod 600.
verify_appliance_demo6 of 7 parameters describedExecutes an automated 5-point demo verification check (API Session, License, CUBEs, Phones, Call Records) certifying demo readiness.
build_stig_template7 of 7 parameters describedPlan or apply a RHEL 8.10 DISA STIG Proxmox template build on proxmox-minis-lab. Claim first (6h bare claim, Ready before Packer), pinned SSG 0.1.82 / DISA V2R8 benchmark / Packer 1.16.0, both scans must be fail=0, template verified, then the claim renews to the TTL cap. Apply needs the plan's planDigest.
create_stig_claim7 of 11 parameters describedPlan or apply a claim-backed clone of the newest STIG template: thin-pool guard (<=70%), one re-encrypting clone at a time, fresh tpmstate0, first-boot LUKS re-key bound to the clone's TPM2, recovery passphrase escrowed to Doppler by name, verify, fstrim. Apply needs the plan's planDigest.
validate_stig_claim2 of 3 parameters describedRun the pinned, tailored oscap STIG scan on a STIG clone, gate on fail=0, record the LUKS volume-key digest, and annotate the claim with the secret-free receipt summary.
get_stig_receipt2 of 2 parameters describedRead a STIG claim's descriptor and receipt summary (counts, fail=0 gate, versions, artifact sha256, key digest) and what status.stig will carry when the operator projection is enabled.
release_stig_template2 of 3 parameters describedRelease a STIG template: verify it is the claimed template, destroy it explicitly (the operator never reaps templates), then delete its claim. Requires confirmDestroy=true.
publish_stig_template6 of 6 parameters describedPlan or apply publishing a sealed RHEL 8.10 STIG template (on the build node) as a compressed qcow2 on the shared image store ct-stig-images: export from the read-only base LV, template.json (hardware, scan counts, digests), sha256, detached SHA-512 signatures with the pinned evidence key (Doppler into an ephemeral GNUPGHOME), index.json (latest, keep 3; prune never touches latest or an image a node is importing). Idempotent: a no-op when the image already verifies. Apply needs the plan's planDigest.
sync_stig_template4 of 4 parameters describedPlan or apply materializing a published STIG template on one node (minis, proxmox, prox-ibm64): claim first (tpl-<template>[-<node>]), verify on the node (pinned key fingerprint, index.json, template.json and qcow2 sha256 + signatures; refuses and releases the claim on any mismatch), qm create with import-from onto thin storage (lvmthin only, pool <= 70%), efidisk, cloud-init, qm template, claim renewed to the TTL cap, sync receipt. Idempotent. Apply needs the plan's planDigest.
collect_stig_evidence17 of 21 parameters describedPlan or apply the DoD evidence package (E1-E10) for a claimed STIG clone, mirroring ct-meta evidence.sh: clone-side stages through the claim's guest access (DISA/SSG scan results, Red Hat OVAL, volume-key digest, rpm inventory and host facts, FIPS state, verification), then the skill's own generators (CKL, POA&M, deviations, PPSM, SBOM, crypto modules, applicability, report, explorer) and the signed, verified bundle. Read-only on the guest and on golden artifacts; output goes to a scratch plan root unless one is named; returns a content-free receipt (counts, sha256, verify result). Apply needs the plan's planDigest.
Tested in a sandbox with an empty environment: no API keys, a throwaway folder, nothing from a real account. A passing test shows the server runs and responds. It is not a security audit.