Receber por email
← Voltar para o catálogo

ThreatLocker Approval Requests

Use this skill when triaging ThreatLocker application approval requests — the heart of day-to-day ThreatLocker operations. Covers pulling the pending queue, grouping requests by application/hash, applying signed-publisher heuristics, and recommending approve/deny decisions with audit-friendly reasoning.

31estrelas
Atualizado há 4 meses

Ver no GitHub ↗ Licença: Apache-2.0

Como adicionar

/plugin marketplace add wyre-technology/msp-claude-plugins

O comando exato pode variar conforme o repositório. Confira o README no GitHub.

Para o autor da skill

Cole no README do seu repo

Mostra que sua skill está catalogada na Skillteca, gera backlink e tráfego rastreável.

[![Skillteca](https://www.skillteca.com.br/api/badge/threatlocker-approval-requests/svg)](https://www.skillteca.com.br/skills/threatlocker-approval-requests)

Skills relacionadas

Ver todas de Documentos →

Alerta por categoria

Receba novas skills de Documentos toda segunda

1 email curto, só com as skills novas de Documentos. 4 minutos de leitura, sem spam, cancela com 1 clique.

Você confirma o email no primeiro envio. Sem spam. Cancela com 1 clique.

Ler descrição completa

ThreatLocker Approval Requests

Approval requests are how ThreatLocker users ask for an application that's currently being blocked to be permitted. Each request carries rich context — file path, hash, signer, the user who asked, the computer it ran on, and a free-text justification. Triaging this queue well is the difference between a productive ThreatLocker deployment and a frustrated client.

API Tools

List Approval Requests

threatlocker_approvals_list

POST-based GetByParameters — see api-patterns. Useful filters via searchText: file name, application name, requester user, computer name. Each request typically includes:

  • requestId, applicationName, fileName, filePath, fileHash
  • signer (Authenticode publisher), signerVerified boolean
  • requesterUser, computerName, computerId, organizationName
  • justification (free-text from the user), dateRequested
  • status — Pending, Approved, or Denied

Get Approval Request

threatlocker_approvals_get

Full detail for one request — includes additional context like the specific policy that blocked it and any prior history of the same hash.

Pending Count

threatlocker_approvals_pending_count

Cheap call — returns the size of the pending queue. Use this as the opening move every shift to know whether you have 3 requests or 300.

Get Permit Application Detail

threatlocker_approvals_get_permit_application

Returns the application that would be permitted if the request is approved — including which group(s) the approval would scope to and what other endpoints would be affected. Always check this before approving anything that looks like it could have a wider blast radius than expected.

Common Workflows

Daily Queue Triage

  1. threatlocker_approvals_pending_count — sets expectations.
  2. threatlocker_approvals_list with status: "Pending",

orderBy: "dateRequested", isAscending: true (oldest first).

  1. Group results by fileHash — many requests are duplicates of the

same binary asked for from multiple endpoints. Decide once, apply broadly.

  1. Within each hash, classify (see heuristics below).
  2. For approves, call threatlocker_approvals_get_permit_application

to confirm scope. Then approve.

  1. For denies, capture a reason that the requesting user can act on.

Signed-Publisher Heuristics

Signals that increase confidence to approve:

  • signerVerified: true AND signer is a known vendor (Microsoft,

Adobe, JetBrains, Mozilla, Google, etc.).

  • filePath is in a vendor-installed location (C:\Program Files\...).
  • The same hash has been previously approved for other endpoints.
  • The requester user matches the computer's primary user.
  • Justification includes a specific business reason ("Need Wireshark

for ticket INC-5123").

Signals that warrant needs-review or deny:

  • Unsigned or signer-unverified binary.
  • Path is in %TEMP%, %APPDATA%, Downloads, or C:\Users\Public.
  • File name mimics a system binary (svchost, lsass, runtime)

but path is unusual.

  • Justification is empty, generic ("need this"), or copy-pasted across

many requests.

  • Hash has been denied previously, especially with a security reason.

Hard escalation triggers — surface to a senior analyst:

  • Known LOLBin (certutil, mshta, bitsadmin) requested from a

user-writable path.

  • RAT/remote-tool installer (AnyDesk, ConnectWise Control,

ScreenConnect) requested by an end user rather than IT.

  • Phishing dropper indicators — Office macros, ISO-mounted shortcuts,

HTA files.

Bulk Approve Pattern

When a single hash appears across many requests:

  1. Pick one representative request and review fully.
  2. Approve via the application — the resulting policy will cover all

endpoints in the chosen group.

  1. The other pending requests for the same hash typically resolve

automatically once the application is permitted at group scope.

  1. Re-pull pending after a minute and confirm the duplicates cleared.

Edge Cases

  • Vendor binaries with broken signatures — Newer versions of some

legitimate apps occasionally ship with signature timing issues. Verify the hash with VirusTotal or the vendor before approving an unsigned binary that should be signed.

  • Self-signed installers — Common in line-of-business apps.

Approve on hash, not signer.

  • Stale requests — A request older than 30 days where the user has

since left the org should usually be denied with a "stale request" reason.

  • Status transitions — A request in Approved/Denied is

terminal; do not attempt to re-process it. Filter to Pending only.

Best Practices

  • Always document a one-line reason on every decision — the audit log

is your defense if a policy change ever needs review.

  • Approve at the application level (hash + signer), not file path —

paths drift, hashes don't.

  • Re-check the queue immediately after a bulk approve to confirm

duplicate requests resolved.

  • For unfamiliar binaries, check the audit log (audit-log skill) to

see what the binary actually did before deciding.

Related Skills

  • [api-patterns](../api-patterns/SKILL.md) — Auth and pagination
  • [audit-log](../audit-log/SKILL.md) — Behavior of binaries before

approval decisions

  • [computers](../computers/SKILL.md) — Endpoint context for requests
  • [computer-groups](../computer-groups/SKILL.md) — Policy scope of

the resulting permit

Compartilhar X LinkedIn