How can we help you?

Agenticify uses GitHub OAuth for authentication. To create an account:

1. Click Sign in with GitHub on the login page
2. Authorize Agenticify to access your GitHub account
3. Complete the setup wizard: choose a plan and install the GitHub App

Your GitHub username becomes your Agenticify identity. No separate password needed.

The GitHub App is required for Agenticify to access your repositories and post review comments. To install:

1. After signing in, you'll see an "Install GitHub App" prompt on the dashboard
2. Click the button — you'll be redirected to GitHub
3. Choose your personal account or organization
4. Select All repositories or pick specific ones
5. Click Install

After installation, go to Repositories and click Sync from GitHub to import your repos.

1. Navigate to Repositories
2. Toggle the checkbox next to the repos you want to activate
3. Click Save

For per-repo customization, click Edit on any repository to configure individual feature toggles (AI review, convention check, risk score) and custom prompts.

Admins are team members who can access and manage your Agenticify account. They see the same dashboard, repositories, and settings as you.

1. Go to Admins from the sidebar
2. Enter the GitHub username of the person you want to invite
3. Click Invite

The invited user must have an existing Agenticify account (signed in with GitHub at least once). They can then switch to your account using the account switcher in the navigation.

When a pull request is opened or updated on an active repository, Agenticify automatically reviews it through a multi-stage pipeline:

Reviews typically complete within 1–2 minutes. Each review can run up to 4 workflow checks and 2 review delivery methods depending on your configuration:

Workflow Checks (appear as check runs on the PR):

• AI Review — General code quality, bugs, security analysis
• Code Convention Check — Pass/fail against your defined rules
• Branch & Commit Check — Naming convention validation
• PR Risk Score — 0–100 risk assessment with approval recommendation

Review Delivery (how feedback is posted):

• PR Comment Review — A consolidated comment on the PR conversation
• Inline PR Review — Line-by-line comments with approve/request changes action

Each feature is independently controlled at both the account and repository level. A feature only runs when both toggles are ON.

The Code Convention Check creates a pass/fail check run on your PR. The AI evaluates your code against the convention rules you define.

You can configure rules at two levels:

• Account level — In Settings, set default convention rules for all repos
• Repository level — In Repositories > Edit, override with repo-specific rules

Example rules: "Use PSR-12 formatting, prefer named imports, handle errors explicitly."

If a repo is set to "Use account defaults", it inherits the account-level rules.

The Risk Score is a 0–100 score that evaluates how risky a pull request is. It considers these factors:

• File count (0–20 pts) — More files changed = higher risk
• Lines changed (0–25 pts) — Larger diffs = higher risk
• Critical files (0–30 pts) — Auth, payments, config, database, deploy files
• Complexity (0–15 pts) — Nested loops, eval(), superglobals, etc.
• AI-detected issues (0–30 pts) — 10 pts per critical issue, 3 pts per warning

Based on the score, Agenticify creates a check run with an approval recommendation:

• Low (0–25) — Safe to merge
• Medium (26–50) — Review recommended
• High (51–75) — Thorough review required
• Critical (76–100) — Do not merge without careful review

Approval is never recommended when critical issues are found, regardless of the overall score.

Yes. In Settings you can customize three types of prompts:

• AI Review Prompt — Guides the overall code review (what to focus on, what to ignore)
• Code Convention Rules — Defines the pass/fail convention rules
• Branch & Commit Convention Rules — Defines naming patterns for branches and commit messages

Each repo can also have its own override prompt (Edit repository > uncheck "Use account defaults").

Before running any checks, Agenticify verifies eligibility. A review is skipped if any of these conditions are true:

• No Subscription — Your account has no active subscription
• Subscription Expired — Your trial or subscription has ended
• Repo Not Found — The repository is not registered in Agenticify
• Repo Inactive — The repository is not activated (toggle is off)
• Quota Exhausted — You've used all your monthly reviews

Additionally, a review completes immediately without running any checks if all applicable features are disabled:

• AI Review is off AND
• Code Convention Check is off (or no convention rules defined) AND
• Branch & Commit Check is off (or no convention rules defined)

You can see the skip reason on the review detail page in your dashboard.

Each feature has a two-level toggle: account-level (in Settings) and repository-level (in Edit Repository). A feature only runs when both are enabled.

Workflow Checks:

• AI Review — When enabled, the AI analyzes your diff for bugs, security issues, and code quality
• Code Convention Check — Only runs when enabled AND convention rules are defined. Creates a pass/fail check run
• Branch & Commit Check — Only runs when enabled AND branch/commit convention rules are defined
• PR Risk Score — Calculates risk based on structural factors + AI-detected issues

Review Delivery:

• PR Comment Review — Posts a consolidated summary comment on the PR. Updated on re-reviews instead of creating duplicates
• Inline PR Review — Posts line-by-line comments and submits a formal review action (approve, request changes, or comment)

Important interactions:

• PR Comment and Inline Review both require AI Review to be enabled (they deliver AI results)
• Convention checks require rules to be defined — enabling the toggle without rules has no effect
• Risk Score uses AI-detected issues in its calculation. If AI Review is disabled, the score is based only on structural factors (file count, lines changed, critical files)
• Deep Context enhances AI Review quality but has no effect if AI Review is disabled

When Inline PR Review is enabled, Agenticify submits a formal GitHub review with one of three actions:

• APPROVE — When the AI finds no issues in your code
• REQUEST_CHANGES — When critical issues are detected
• COMMENT — When only warnings or suggestions are found

Inline comments are placed on specific lines of the diff. Only added lines (lines starting with + in the diff) can receive inline comments. Issues on deleted or unchanged lines are included in the review body instead.

On re-reviews (when you push new commits), previous inline comments from Agenticify are automatically removed and replaced with fresh ones.

Deep Context pre-scans your repository to build a structured codebase profile, then injects this context into every AI review. This makes reviews significantly more accurate and architecture-aware.

When enabled, the AI understands:

• Your project's architecture (MVC, microservices, etc.)
• Tech stack — languages, frameworks, and package managers
• External services (databases, caches, APIs)
• Directory structure and key file locations
• Naming conventions and coding patterns

In addition, the AI receives the full file contents of changed files (not just the diff), allowing it to understand the complete context of each change.

Without Deep Context, the AI only sees the PR diff — it has no knowledge of the broader codebase. With Deep Context:

• Better pattern recognition — The AI knows your naming conventions and can flag inconsistencies
• Architecture-aware feedback — It understands where code should live and suggests proper placement
• Dependency awareness — It knows which services and databases your project uses
• Full file context — The AI sees the complete file, not just changed lines, catching issues that span the entire file

Deep Context is generated automatically when you activate a repository for the first time. You can also manually regenerate it at any time:

1. Go to Repositories and click Edit on any repo
2. In the sidebar, find the Deep Context card
3. Click Regenerate Context

We recommend regenerating context when your project undergoes major structural changes (new frameworks, directory reorganization, etc.).

Yes. Deep Context can be controlled at two levels:

• Account level — The Deep Context master toggle in Settings controls it for all repos
• Repository level — Each repo has its own Deep Context toggle in Edit Repository (sidebar)

Both toggles must be ON for Deep Context to be used in reviews. When disabled, reviews revert to stateless mode (diff-only analysis).

Agenticify integrates with Jira Cloud via OAuth 2.0:

1. Go to Integrations
2. Click Connect Jira
3. You'll be redirected to Atlassian — authorize Agenticify to access your Jira instance
4. After authorization, your Jira connection appears on the Integrations page

Once connected, you'll see your Webhook URL and can optionally set a Webhook Secret for signature verification.

The webhook tells Agenticify when Jira tickets are created or updated:

1. In your Jira instance, go to Settings > System > WebHooks
2. Click Create a WebHook
3. Paste the Webhook URL shown on the Agenticify Integrations page
4. Under Events, select: Issue created and Issue updated
5. Save

For extra security, enter a Webhook Secret on both sides. Agenticify will verify the HMAC-SHA256 signature on every incoming webhook.

Mappings link your Jira projects to GitHub repositories so Agenticify knows which repos belong to which project.

1. Go to Integrations > Manage Mappings
2. Your Jira projects are listed automatically
3. Select the checkbox next to a project and click Select to choose repositories
4. A project can be linked to multiple repositories

Mappings are used for both task readiness checks (posting comments to the right connection) and agentic development (knowing which repo to code in).

When a Jira project is linked to multiple repositories, component mappings help Agenticify automatically determine which repo a ticket belongs to.

1. In the mapping modal, click Edit Repos on an existing mapping
2. When 2 or more repos are selected, the Component Mapping section appears
3. Click Add Component
4. Enter the Jira component name (must match exactly) and select the target repo

Example: Jira project "PLATFORM" has repos api-service and web-app. Map component "Backend" → api-service and "Frontend" → web-app.

When an agentic task is triggered, the ticket's component is used to auto-select the correct repository.

When a Jira ticket is created or changes status, Agenticify's AI evaluates its readiness for development. This ensures no vague or incomplete tasks make it to development:

Scoring criteria:

• Clarity (25%) — Is the ticket clearly written?
• Acceptance Criteria (30%) — Are success conditions defined?
• Technical Details (25%) — Are implementation details specified?
• Scope (20%) — Is the scope reasonable for a single task?

Results:

• Ready — Ticket is clear enough for development
• Needs Improvement — Some criteria not met, suggestions given
• Not Ready — Too vague or incomplete

You can customize the readiness criteria in Settings > Task Readiness Prompt.

Agentic Development is Agenticify's AI coding capability. An AI agent reads a validated Jira ticket and writes the code, creates a branch, and opens a pull request — all automatically.

The full flow:

Agentic Development is available on Business and Enterprise plans.

Tasks are triggered when two conditions are met simultaneously on a Jira ticket:

1. The ticket has a label named "agenticify" (case-insensitive)
2. The ticket's status changes to "In Progress" (or similar: "In Development", "In Dev")

Prerequisites:

• Jira must be connected with webhook configured
• The Jira project must be mapped to at least one repository
• Your subscription must have agentic task quota remaining

You can also still create tasks manually from the Task Readiness page using the Start Task button.

When a task is triggered, Agenticify needs to know which repository the AI should code in. It resolves the repo in this order:

1. Component mapping — If the Jira ticket has a component and that component is mapped to a repo, use that repo
2. Single repo — If the Jira project is mapped to only one repository, use it automatically
3. Can't determine — If multiple repos are mapped and no component matches, the task requires manual approval with repo selection

Set up component mappings in Integrations > Manage Mappings > Edit Repos to enable automatic resolution.

Agenticify uses a smart approval system to decide whether a task should start immediately or wait for your review:

Auto-approved (starts immediately) when:

• "Auto-approve Agentic Tasks" is ON in Settings, AND
• The repository is clearly resolved (via component or single-repo mapping)

Requires approval when any of these are true:

• "Auto-approve Agentic Tasks" is OFF — all tasks need approval
• Repository can't be automatically determined

When a task needs approval:

1. A warning banner appears on your dashboard showing the count
2. Click Review Tasks to see all pending tasks
3. On the task detail page, you can:
   • Approve & Start — optionally select a repository, then the AI begins
   • Reject — marks the task as failed

An agentic task goes through these stages:

• Awaiting Approval — Waiting for you to review and approve
• Pending — Approved, queued for processing
• Cloning — AI is cloning the repository
• Coding — AI is writing code using Claude
• Pushing — Committing and pushing to GitHub
• In Review — PR created, going through code review
• Completed — PR created successfully
• Failed — Something went wrong (see error message)

You can filter tasks by status on the Agentic Tasks page.

Feature toggles in Settings are master switches that control features across your entire account. They are organized into groups:

Review Delivery — How review feedback is posted on your PRs:

• PR Comment Review — Consolidated review comment on the PR conversation
• Inline PR Review — Line-by-line comments with approve/request changes

Workflow Checks — Automated checks that run on your PRs:

• AI Review — AI-powered code quality, security, and bug analysis
• Code Convention Check — Pass/fail check run for coding standards
• Branch & Commit Check — Validates branch names and commit messages
• PR Risk Score — Risk analysis with approval recommendation

Deep Context — Codebase-aware AI reviews with architecture understanding

Jira & Agentic (visible when Jira connected):

• Task Readiness — AI-powered readiness checks for Jira tickets
• Auto-approve Agentic Tasks — Whether tasks start automatically or wait for approval

When a master toggle is OFF, that feature is disabled everywhere — even if individual repos have it enabled. A feature only runs when both the account-level and repo-level toggles are ON.

Agenticify uses a two-level configuration system for both feature toggles and prompts:

Feature toggles (on/off switches):

• Account level (Settings) — Master switches. When OFF, the feature is disabled for all repos.
• Repository level (Edit Repository) — Per-repo switches. Only effective when the account toggle is also ON.

Both must be ON for a feature to run. If the account toggle is OFF, repo-level toggles show a Disabled at account level badge.

Prompts and rules:

• Account defaults (Settings) — AI review prompt, code convention rules, branch & commit rules. Applied to all repos that use defaults.
• Repository overrides (Edit Repository) — Each repo can switch between "Use default" and "Use custom" for each prompt type.

Important: Convention checks (code and branch/commit) require rules to be defined. If you enable the toggle but don't write any rules (at either level), the check simply won't run.

This toggle controls whether agentic tasks start automatically or require your manual approval:

• ON (default) — Tasks with a clearly resolved repository start immediately. Tasks where the repo can't be determined still require approval.
• OFF — All agentic tasks require manual approval before the AI begins coding. Use this if you want to review every task before it starts.

This setting is in Settings, under the Jira section (only visible when Jira is connected).

Agenticify offers three plans:

• Team — $49/month
  PR reviews, code conventions, risk score. Ideal for small teams.
• Business — $199/month
  Everything in Team + Jira integration, task readiness checks, and 100 agentic tasks/month. For teams using Jira.
• Enterprise — $799/month
  Everything in Business + 300 agentic tasks/month, SSO, and priority support. For large organizations.

Every new account gets a 7-day free trial:

• Full access to all features
• No credit card required to start
• A trial banner shows your remaining days

When the trial ends, you need to choose a plan to continue using Agenticify. Your data is preserved.

Each plan has monthly usage limits that reset at the start of each billing period:

• Reviews — Number of PR reviews per month
• Readiness Checks — Number of task readiness checks per month
• Agentic Tasks — Number of AI coding tasks per month (Business & Enterprise only)
• Repositories — Number of active repositories

You can see your current usage on the Dashboard and Billing pages.

Yes, you can cancel anytime from your Billing page.

• You keep access until the end of your current billing period
• No refunds for partial periods
• You can resume your subscription before it ends

Your payment is handled securely by Stripe:

1. Go to Billing
2. Click Payment Settings
3. Update your card details in the Stripe customer portal