Skip to content

Vault Integration - Inject secrets from AWS Secrets Manager / HashiCorp Vault without exposing them in scrollback or shell history #9878

@networkbm

Description

@networkbm

Pre-submit Checks

Describe the solution you'd like?

When working in regulated or security-sensitive environments (FedRAMP, CMMC, DoD IL4/IL5), developers and security engineers frequently need to inject secrets into their shell session. The current workflow requires exporting secrets as environment variables directly in the terminal:

export AWS_SECRET_ACCESS_KEY=EXAMPLE

This exposes secrets in shell history, Warp's scrollback buffer, and any session logs. In compliance-controlled environments this is a policy violation and a real audit finding. Warp's agent-first model makes this worse — agents executing commands on the user's behalf increase the surface area for
unintended secret exposure.

Proposed solution: A warp vault inject command that fetches a secret from a secrets manager and injects it as an environment variable in memory only - never written to scrollback, history, or logs.

warp vault inject aws/prod/api-key --as API_KEY

Scrollback redaction: if a secret value appears in terminal output anyway, Warp masks it with [REDACTED].

Scope for v1:

  • AWS Secrets Manager support
  • In-memory injection only — no shell history writes
  • Scrollback redaction for known secret values
  • Config file for mapping vault paths to env var names

Future providers: HashiCorp Vault, 1Password, Azure Key Vault

Prior art:

  • direnv — no secrets manager integration, no redaction
  • chamber — external to the terminal, no scrollback protection
  • 1Password CLI (op run) — closest analog, but not terminal-native

Is your feature request related to a problem? Please describe.

No response

Additional context

No response

Operating system (OS)

Select an OS

How important is this feature to you?

1 (Not too important)

Warp Internal (ignore) - linear-label:39cc6478-1249-4ee7-950b-c428edfeecd1

None

Metadata

Metadata

Assignees

No one assigned

    Labels

    area:agentAgent workflows, conversations, prompts, cloud mode, and AI-specific UI.area:shell-terminalTerminal input/output, shell integration, prompt behavior, and block rendering.enhancementNew feature or request.repro:highThe report includes enough evidence that the issue appears highly reproducible.triagedIssue has received an initial automated triage pass.

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions