Skip to content

Usage Credits

Fencer bills its AI actions in credits. AI fixes, AI PR reviews, AI codebase reviews and AI penetration tests all draw from one balance, so you decide where your AI budget goes. Your plan includes credits, and admins can buy more at any time.

Credits replace the per-action limits (the AI Pentests, AI Fixers and AI Reviews meters, and AI pentest credits counted per run). Each organization moves to credits on its own switch date; until then, the old limits described in Billing and Usage still apply. See Switching to credits.

  • An organization that has moved to credits. Its Billing page under Admin then has a Credits tab.
  • Buying credits needs permission to edit organization settings (the Admin role). Anyone else is told which admin to ask.
  • Credit history on the Credits tab needs permission to read organization settings. To see the purchases and grants behind the Extra or Extra, expiring figure, select the list icon on that card. Each row shows the source, the credits left of the original amount and, for expiring credits, the expiry date. Only users who can read organization settings see this icon.

Every button that starts a priced action shows its cost. Hover over the cost to see your balance and the rule that applies.

Action Cost
AI fix 1 credit covers the fix and 2 follow-ups you ask for
Bucket fix 1 credit per pull request
AI PR review 1 credit covers 3 completed review rounds on a pull request
AI codebase review 5, 7 or 10 credits per repository, by lines in scope
AI penetration test, full or scoped 3,000 credits, charged when the test reports a new validated High or Critical finding

Failed actions aren’t charged. Questions to the Fencer agent never use credits; see Questions to the Fencer agent.

  • One credit covers 3 rounds of one fix: the fix itself and the next 2 follow-ups that people ask for in a comment or review on its pull request. The next follow-up after that uses another credit and covers 3 more rounds.
  • Only follow-ups people ask for count. Fencer also answers CI failures, merge conflicts and other bots’ comments on its pull requests, and answers at most 3 rounds of bot comments per fix before asking a person to comment instead.
  • A round that fails, or that commits nothing to the pull request, isn’t counted.
  • A bucket fix, which fixes a group of vulnerabilities in one pull request, costs 1 credit per pull request.

Follow-ups run while the Autofix checkbox in the pull request description is ticked, which it is by default. Untick it to stop Fencer from answering feedback on that pull request. See Automated Remediation.

  • One credit covers 3 completed review rounds on a pull request. A round is a completed review of a new commit, so a push, a rebase or a force-push that changes the head of the pull request starts a new round.
  • Only completed reviews of a new commit count. A review that fails, or that is discarded because the pull request changed while it ran, isn’t counted.

The cost depends on the lines of code left in scope after your Scope and Ignore settings:

Lines in scope Cost
Up to 25,000 5 credits
Up to 100,000 7 credits
More than 100,000 10 credits

The estimate shows the cost before you click Start review. A completed review is charged even when it finds nothing, and an empty repository costs 5 credits.

  • A full or scoped penetration test costs 3,000 credits. Starting a run holds them, so you need 3,000 available to start. A scheduled test holds them when its run starts, not when you schedule it.
  • The credits are charged when the full test finishes and reports a new validated High or Critical finding; otherwise they’re released. A finding that an earlier charged test of your organization already reported doesn’t count again.
  • Stop pentest ends a run and hides the findings that run reported first.
  • If a run times out, start it again to complete the test. It keeps the findings it already reported.

Questions you ask the Fencer agent, in the web chat or in Slack, never use credits. Each plan includes a number of questions every month:

Plan Questions per month
Basic 20
Essentials 30 per contributor
Growth 60 per contributor
Advanced 120 per contributor
Penetration Testing Only Not included

When your organization reaches its limit, the agent pauses until the next billing period. Extra questions can’t be bought: more contributors or a higher plan raises the allowance.

Plan Included credits
Basic 10 a month
Essentials 15 per contributor a month
Growth 30 per contributor a month
Advanced 60 per contributor a month
Penetration Testing Only 3,000 a year, billed yearly
  • Trials get half of their plan’s credits.
  • Included credits reset at the start of each billing period. Unused included credits don’t carry over.
  • Included credits are always used first, before credits you bought.

Credits on the Penetration Testing Only plan cover AI penetration tests only, full and scoped. Other AI actions, such as AI fixes, AI PR reviews and AI codebase reviews, aren’t included: they’re hidden, and where one is reached the app says This action is not included in your plan.

Open Billing under Admin and select the Credits tab.

  • The balance shows the credits available now, split into Included credits (with the date they reset) and Extra credits you bought, which never expire. Credits held by running actions, such as a penetration test in progress, are shown separately.
  • The list icon on the Extra or Extra, expiring card lists each purchase and grant, with what is left of it and when it expires, if ever.
  • Credit history lists every action that held, used or released credits, with who started it and what it was for.

Credits cost $1 each, and credits you buy never expire.

  1. Open Billing under Admin, or click Buy credits wherever Fencer says you are out of credits.
  2. Choose how many credits to buy.
  3. Complete the checkout. The credits are added to your balance.

During a trial, Buy credits opens the Plans tab instead, so you can choose a plan.

Only users who can edit organization settings can buy credits; everyone else is told which admin to ask. There is no separate AI penetration test purchase: a penetration test costs 3,000 credits from your balance.

When your balance can’t cover an action:

  • Automatic actions pause. Automatic AI PR reviews and fix follow-ups are skipped, and nothing is charged. They resume once credits are available again, after a purchase or when your included credits reset.
  • Actions you start show Out of credits instead of starting, with Buy credits for admins.
  • Organization admins are notified when the balance runs out.

Fencer never charges you beyond your balance automatically.

Organizations move from the per-action limits to credits at different times:

  • Each organization moves at the end of its current billing period.
  • Organizations without a subscription move immediately.
  • Organizations on an annual subscription move with the next monthly billing cycle.
  • Admins see a banner on Billing with the date their organization moves.

At the switch:

  • Each unused AI pentest credit converts to 3,000 credits.
  • Work already running when your organization switches, such as a penetration test in progress, finishes under the old rules.
  • New organizations start on credits from the day they sign up.

Until your organization’s switch date, the AI meters and pentest credits described in Billing and Usage apply.

  • There is no Credits tab — Your organization has not switched to credits yet. Check the date on the banner on Billing.
  • A button shows “Out of credits” — Your balance can’t cover the action. An admin can buy credits, or wait for your included credits to reset.
  • “This action is not included in your plan” — Your plan doesn’t include this AI action. On Penetration Testing Only, credits cover AI penetration tests only.
  • A penetration test won’t start — Starting a test needs 3,000 credits available.
  • An automatic review didn’t run — Automatic reviews pause at zero. Buy credits, then re-run the review from the pull request comment.
  • The Fencer agent stopped answering questions — Your organization has reached its monthly question allowance. The agent answers again from the next billing period; more contributors or a higher plan raises the allowance.