Warp

Contradictions

Two pages in the same docs that state incompatible facts. Each card quotes both sides verbatim and links back to the live page.

1. Minimum files indexed per codebase

   The docs promise a 5,000-file floor on every plan; the pricing table caps Free at 3,000, so a Free user hits "Codebase too large" well before the documented minimum.

   agents/capabilities/codebase-context

   All plans support indexing at least 5,000 files per codebase, with higher tiers including support for more files and additional codebases.

   pricing

   | Files per codebase | 3,000 | 100,000 | 100,000 | 100,000 | Custom |

2. Who is billed for scheduled cloud agent runs

   Either an unattended run drains the team pool or it drains one person's plan credits first, and the two pages cannot both be the billing rule.

   platform/triggers/scheduled-agents

   All usage is billed to the team’s shared credit balance.

   support-and-community/plans-and-billing/pricing-faqs

   Some cloud agent runs aren’t initiated by a specific team member — for example, scheduled runs or runs triggered through an agent API key. On self-serve plans (Build, Max, Business), these runs are billed to the team owner.

3. Which add-on pool pays for agent-API-key runs

   The second bucket is the owner's personal add-on balance on the API keys page and the team-wide shared pool on two other pages.

   reference/cli/api-keys

   On Build, Max, and Business plans, bill the team owner: the owner’s plan-included credits, then the owner’s add-on credits.

   platform/team-access-billing-and-identity

   Warp bills the team owner. The waterfall is: the owner’s plan-included credits, then the team’s shared add-on credit pool.

4. First bucket in the user-triggered credit waterfall

   One section starts the "fixed order" with the user's cloud agent credits, the other starts it with the user's plan-included credits, for the same user-triggered run.

   platform/team-access-billing-and-identity

   User-triggered runs on Build, Max, or Business \- Warp draws from any cloud agent credits the user has, then the user’s plan-included credits, then the team’s shared add-on credit pool.

   platform/team-access-billing-and-identity

   On Build, Max, and Business plans, Warp consumes credits in a fixed order. You cannot prioritize user-scoped grants ahead of shared team grants:
   1. The triggering user’s monthly credits included with their plan.

5. Whether Warp may train on customer data

   One page reserves the right to train on collected data; the docs home states flatly that none of it is used for training.

   agents/getting-started/faqs

   Warp reserves the right to use data collected to train models and improve Warp.

   Warp is SOC 2 compliant and has Zero Data Retention policies with all contracted LLM providers. No customer AI data is retained, stored, or used for training.

6. Is Secret Redaction on by default or opt-in

   The privacy page and security overview both describe redaction as always on, while the feature page says you have to switch it on; the security overview repeats the automatic claim for Enterprise.

   support-and-community/privacy-and-security/privacy

   Warp unconditionally applies Secret Redaction in all AI interactions to ensure that any sensitive data is _never_ collected or sent to third parties.

   support-and-community/privacy-and-security/secret-redaction

   Disabled by default, to enable Secret Redaction open Settings > Privacy > Secret redaction or type in "Secret Redaction" to toggle it in the Command Palette.

7. Can Free users disable telemetry and keep AI

   The FAQ tells any user they can turn telemetry off and keep full ZDR; the privacy page says a Free user who does that loses AI entirely.

   support-and-community/plans-and-billing/pricing-faqs

   Individual level — Any user can enable full ZDR for their own account by disabling Help Improve Warp in Settings \> Privacy.

   support-and-community/privacy-and-security/privacy

   Telemetry must be enabled to use AI features on the Free plan, while paid plans can opt out at any time and continue using Warp, including AI.

8. Third-party harnesses on the Free plan

   The pricing page sells harness choice as a Free-plan feature while the platform docs gate it behind Build; the docs pricing FAQ adds a third reading by calling the beta "available to all users".

   pricing

   Use any harness in the cloud (beta)

   platform/harnesses

   Third-party harnesses require a Build plan or higher. On the Free plan, cloud agent runs use Warp Agent, and choosing another harness returns an upgrade prompt.

9. Image attachments on cloud agent runs

   The CLI reference documents an image flag on run-cloud with a hard limit, and the platform FAQ says cloud runs take no images at all.

   reference/cli

   --attach <PATH> — attach an image file to the agent query. Can be repeated (maximum 5).

   platform/faqs

   Cloud agent conversations do not currently support image attachments.

10. Singular vs plural path for creating a run

    The generated API reference publishes two paths for the same create-run operation with identical descriptions and no canonical marker, so an integrator has to guess.

    api

    ### POST /agent/run
    
    Run an agent task
    
    Spawn a cloud agent with a prompt and optional configuration.
    The agent will be queued for execution and assigned a unique run ID.

    api

    ### POST /agent/runs
    
    Run a cloud agent
    
    Spawn a cloud agent with a prompt and optional configuration.
    The agent will be queued for execution and assigned a unique run ID.

11. Query parameter for filtering runs by config name

    One page names two different parameters for the same filter; the generated API reference documents name, so anything written against config_name will not filter.

    reference/api-and-sdk

    Use the name query parameter on GET /agent/runs to filter runs by config name.

    reference/api-and-sdk

    List runs with pagination and filters for state, config_name, model_id, creator, source, and creation time.

12. What (?i) does to a secret redaction regex

    The same sentence says regexes are case-sensitive by default and that (?i) makes them case-sensitive, so a user cannot tell which behavior to code against.

    support-and-community/privacy-and-security/secret-redaction

    Secret redaction regexes are case-sensitive by default (i.e. the regex password will not match the text Password).

    support-and-community/privacy-and-security/secret-redaction

    If you want a regex to be case-sensitive, you can prepend (?i) like so: (?i)password to ensure that PASSWORD, Password, and password would all match.

13. fish shell support on Windows

    The shells page lists fish as supported on Windows; the known issues page lists it as unsupported there.

    getting-started/supported-shells

    Warp supports bash, zsh, fish, and PowerShell (pwsh) across macOS, Windows, and Linux, loading your login shell by default.

    support-and-community/troubleshooting-and-support/known-issues

    The following features are not supported in Warp on Windows. Track the relevant GitHub issues linked below for any changes:
    
    - cmd.exe or fish shells

14. macOS Ventura auto-update issue: open or fixed

    The update page still warns of an active Ventura auto-update issue that the known issues page says was resolved.

    support-and-community/troubleshooting-and-support/updating-warp

    (Oct 2022): There is a known issue with auto-update on macOS Ventura.

    support-and-community/troubleshooting-and-support/known-issues

    Warp may have an error opening after auto-update on macOS Ventura. This issue has been resolved for current and future releases of Warp.

15. What happens to add-on credits when a team is deleted

    One page says leftover credits block deletion until they are cleared, the other says deletion leaves them unusable, which changes whether an admin loses paid balance.

    support-and-community/plans-and-billing/add-on-credits

    An admin deletes the team - Any remaining add-on credits must be used or removed before a team can be deleted.

    support-and-community/plans-and-billing/pricing-faqs

    An admin deletes the team: Any remaining add-on credits tied to the team are no longer usable by anyone.

16. Whether the Warp Agent CLI updates itself

    The reference states auto-update as unconditional; the quickstart says Homebrew installs never self-update and must be upgraded by hand.

    agents/cli/reference

    The CLI updates automatically. Installed builds check for updates in the background, download new versions, and apply them the next time you launch warp. A running session is never interrupted.

    agents/cli/quickstart

    Homebrew installations don't update themselves. The CLI still tells you when a new version is available, but you install it by running brew upgrade --cask warp-agent-cli.

17. Secret Redaction described as unconditional next to its own toggle

    The prose calls redaction unconditional while the page's own telemetry table documents an on/off setting for it.

    support-and-community/privacy-and-security/privacy

    Warp unconditionally applies Secret Redaction in all AI interactions to ensure that any sensitive data is _never_ collected or sent to third parties.

    support-and-community/privacy-and-security/privacy

    | Toggle Secret Redaction | Toggled on/off the setting for Secret Redaction - attempts to redact secrets and sensitive information |

18. MCP servers under the default CLI profile

    Following the quickstart with the default profile, MCP tools either load or are unavailable, and the two pages describe the same default run differently.

    reference/cli/quickstart

    This uses the default agent profile, loads any available MCP servers, and executes the run locally.

    reference/cli/agent-profiles

    The agent does not have the ability to use MCP servers by default.

19. Which shell Warp starts on an SSH host

    For the same legacy SSH wrapper, one page says the remote shell is always bash, the other documents zsh sessions with their own bootstrap.

    support-and-community/troubleshooting-and-support/known-issues

    When you SSH, we start a bash shell on the remote host. We built a wrapper around SSH to make Warp features possible.

    terminal/warpify/ssh-legacy

    In remote sessions, only bash or zsh shells are supported. If you're using a different shell, use command ssh directly (see below for more details).

Content issues

Gaps, stale facts, and structural problems that make the docs harder to trust and harder for AI agents to use.

• Stale

  The whole CLI reference section documents a binary whose support ends this month: "oz commands remain supported through the end of September 2026", and "The Warp Agent CLI reference does not yet document warp equivalents for every workflow on this page."

  View page
• Stale

  Cloud Agent FAQs still say "Concurrency limits and credit allocation details are still being finalized" although the pricing page already publishes per-plan concurrency (20 and 80) and credit allowances

  View page
• Structural

  The same product is sold under two names: the pricing page says "Free, Build, Max, and Business users can buy Reload credits for additional usage" while every docs page calls them add-on credits

  View page
• Structural

  The Session Sharing page is now an empty redirect stub, "Session sharing documentation has moved.", yet the pricing FAQ and offline guide still link to it as the feature page for shared terminal sessions

  View page
• Gap

  The security overview's ZDR provider list leaves out Fireworks AI: "Warp has Zero Data Retention (ZDR) agreements with its contracted LLM providers (Anthropic, OpenAI, Google, xAI)", while the model and pricing pages name Fireworks AI as the host for Warp's open-source models

  View page