Agentic Index
cubic vs Graphite (2026)
Both vendors last verified
Two inexpensive ways into AI review, Cubic's Team at $30 per developer and Graphite's Starter at $20 per user, both billed annually, solving different halves of the pull request problem. That verdict is the Agentic Index coverage score, graded from each vendor's own published materials.
Cubic reviews every pull request with whole codebase context, enforces plain English rules and runs background scans that find and fix, free for 20 reviews a month and on public repositories. Graphite restructures the workflow around stacked pull requests with a stack aware merge queue and an agent that reviews, chats, fixes and merges, with a free hobby tier. Cubic makes reviews better, and Graphite makes them smaller, which is arguably the more effective fix.
On the Agentic Index coding agent ranking, neither cubic nor Graphite clears the bar, which asks for all five merge loop capabilities documented in full. cubic does not document observability and auditability in full, nor human oversight and guardrails; Graphite documents two of the five in full. 3 of the 96 vendors in the lane clear it. See the coding agent ranking
This comparison is published by Agentic Index, an independent agentic AI vendor research platform. cubic and Graphite are each graded against the same 14 capability Agentic Index taxonomy, from the vendor's own public materials under the Agentic Index verification standard, alongside 1,084 researched vendors. No vendor pays for placement and no vendor has reviewed this page. How this evidence is graded
Choose cubic if
- Review quality is the immediate need, and you want rules your team writes in plain English.
- Plain English rules are how your team will keep conventions current.
- Background scanning outside the pull request finds what review never sees.
Choose Graphite if
- Your pull requests are too big to review properly and stacking is the structural answer.
- A merge queue that understands stacks removes a real source of friction.
- You want the workflow fixed, not just better commentary on it.
| Feature | c cubic |
G Graphite |
|---|---|---|
| Action & orchestration | ||
|
Integrations & Tool Calling Ability to connect agents to real systems through native integrations, OAuth-authenticated actions, custom tools, APIs, webhooks, or MCP-compatible tools. |
||
|
cubicIntegrations & Tool Calling A two way GitHub app drives inline comments, fix branches and approvals, project connectors cover Jira, Linear, Asana and Notion with Confluence on Pro, Slack and email notifications ship on Pro, PostHog (5 Oct), Sentry (8 Oct) and Datadog (11 Oct 2026) connect as production evidence so scan agents can confirm findings against real errors and logs, and cubic runs inside coding agent hosts through its MCP server and CLI. Sourcecubic.dev/pricing-plans and docs.cubic.dev/changelogread 2026-10-28 |
||
|
GraphiteIntegrations & Tool Calling Deep GitHub integration spans repositories, pull requests, CI and merge, with GitHub Enterprise Server supported on Enterprise. Developers work through a CLI for stacked pull requests and simplified git commands, a VS Code extension, a Mac menu bar app, MCP, a web pull request page and a PR inbox. Slack notifications and a Linear integration ship from Starter, with advanced integrations on Team and above. The CI optimizer integrates Buildkite and GitHub Actions. Cursor Cloud Agents appear inside the pull request for authors on a paid Cursor plan, opening draft pull requests and committing fixes to the branch. The agent acts by commenting inline, applying suggested fixes, generating PR titles and descriptions, summarizing CI failures, restacking and merging through the queue. Sourcegraphite.com/pricing, docs/integrations and docs/agentsread 2026-10-28 |
||
|
Workflow Orchestration Ability to sequence, branch, retry, route, and combine deterministic workflow nodes with autonomous agent steps. |
||
|
cubicWorkflow Orchestration Codebase scans are documented as thousands of parallel agents working through a staged process that maps the codebase, runs parallel checks, investigates, deduplicates and scores findings, and Ultrareview is a documented multi pass analysis. The scan docs still label scans beta and available by request, while the pricing page lists them on Pro. Reviews run a fixed sequence of review, inline comments, fix branches and optional approval, and heavier implementation can be handed to an outside coding agent. Sourcedocs.cubic.dev/codebase-scan/codebase-scans and cubic.dev/pricing-plansread 2026-10-25 |
||
|
GraphiteWorkflow Orchestration A stack aware merge queue sequences approved pull requests to keep the trunk branch green. It comes as Basic on Team and as Advanced, with additional settings, on Enterprise, and an external merge queue integration is in beta. The queue also rebases stacked branches automatically as earlier changes land, so a chain of dependent pull requests stays current without manual intervention. An Automations engine and a CI optimizer batch and selectively trigger test runs, integrated with Buildkite and GitHub Actions. Within the pull request, Graphite Agent chains review into suggested fixes, CI summaries and collaborative merge. Sourcegraphite.com/pricing, docs/graphite-merge-queue and docs/external-merge-queue-integrationread 2026-10-28 |
||
|
Triggers & Channel Coverage How agents wake up and where they work: schedules, webhooks, message events, CRM events, inbox events, chat, email, voice, and collaboration tools. |
||
|
cubicTriggers & Channel Coverage Reviews run automatically when a pull request opens and again on new commits, codebase scans run nightly on Pro, and the AI wiki regenerates on a weekly or monthly schedule, so work reaches the agent without a person starting it. @cubic mentions, Fix with cubic and the CLI add requested runs. Sourcecubic.dev/pricing-plans and docs.cubic.dev/wiki/ai-wikiread 2026-09-29 |
||
|
GraphiteTriggers & Channel Coverage Every pull request gets an automatic review from Graphite Agent when it is opened or updated. Automations act on pull request events to assign reviewers and add labels. Developers also reach the agent from the PR page, Graphite Chat, the CLI, VS Code and MCP. Sourcegraphite.com/docs/ai-reviews, docs/automations and graphite.com/pricingread 2026-10-28 |
||
| Knowledge & context | ||
|
Knowledge Grounding & RAG Ability to ground agent behavior in company data through document ingestion, retrieval, external knowledge APIs, semantic search, or RAG layers. |
||
|
cubicKnowledge Grounding & RAG The AI wiki automatically indexes the codebase into searchable documentation with source links and architecture diagrams, regenerated monthly on Team and on a daily, weekly or monthly schedule on Pro and Max as code changes. It can be queried in plain English, over MCP or as synced markdown, alongside whole repository and cross repository context during reviews. Sourcedocs.cubic.dev/wiki/ai-wiki 2026-09-29 and cubic.dev/pricing-plansread 2026-10-25 |
||
|
GraphiteKnowledge Grounding & RAG Optional code indexing keeps an index of the files in each synced repository and speeds Graphite Chat's tool calls and the PR page; reviews that draw on related files through the index, and faster merges, are marked coming soon, so for now the agent's reviews are not grounded in it. By default Graphite stores no code and reads it from GitHub, and an index that is turned off is deleted within 30 days. Style guides and rules work as instructions to the review agent. Sourcegraphite.com/docs/code-indexing and docs/ai-review-customizationread 2026-10-28 |
||
|
Memory & State Persistence Ability to persist context across a run, conversation, workflow, user, team, or longer-term memory layer. |
||
|
cubicMemory & State Persistence cubic extracts learnings from PR comments, reactions and senior reviewers' past reviews, scoped to the team and never shared across organizations or repositories. Learnings are browsable, editable and deletable under Memory and Learning, and a weekly cleanup merges duplicates and archives outdated guidance. Sourcedocs.cubic.dev/ai-review/memory-and-learningread 2026-09-29 |
||
|
GraphiteMemory & State Persistence Team written custom rules and file based rules persist and apply to every review. Graphite Agent also learns from whether developers accept or reject its comments, though no scope, lifetime or purge path is stated for what it learns. Sourcegraphite.com/docs/ai-review-customization and docs/ai-reviews; graphite.com/docs/ai-reviewsread 2026-10-28 |
||
| Control & trust | ||
|
Human Oversight & Guardrails Approval steps, consent checkpoints, escalation rules, structured guardrails, policy constraints, and pause/resume controls. |
||
|
cubicHuman Oversight & Guardrails Auto approval runs under a per repository policy the customer sets (low risk only, custom or always), and cubic approves only when its review finds no issues. No person approves anything before cubic acts, so the policy constrains the agent and is not an approval step. Shadow mode, which previews approvals before they go live, is covered under testing. Sourcedocs.cubic.dev/ai-review/auto-approvalread 2026-09-29 |
||
|
GraphiteHuman Oversight & Guardrails Review output is advisory. Findings land as inline comments and suggested fixes that a developer chooses to apply. Graphite Chat lets the developer interrogate and refine a suggestion before committing, and the person stays accountable for the merge. Teams constrain the review agent through AI review customization, which covers automations, filters and rules, and through an imported style guide. The merge queue gates what reaches trunk under rules the team configures, with advanced settings on Enterprise, and ACLs on Enterprise limit access. Graphite names no approval gate for each agent action, no permission scoping for the agent itself and no shadow mode. Sourcegraphite.com/pricing, docs/graphite-chat and graphite.com/featuresread 2026-10-28 |
||
|
Security, Identity & Governance RBAC, SSO, auditability, encryption, least-privilege tool access, compliance posture, and data handling policy. |
||
|
cubicSecurity, Identity & Governance Compliance is SOC 2 Type 1, with TLS 1.2 or higher in transit, AES-256 at rest, subprocessor agreements that prohibit training on customer code, and a trust center at trust.cubic.dev listing subprocessors. Enterprise adds SSO and SAML through GitHub OAuth, GitHub Enterprise support, exportable compliance audits and custom MSA and DPA terms. Sourcedocs.cubic.dev/account/privacy-security and cubic.dev/pricing-plansread 2026-10-28 |
||
|
GraphiteSecurity, Identity & Governance The Enterprise plan includes SAML and SSO, access control lists, an audit log with SIEM export, GitHub Enterprise Server support, a custom MSA and private uploads. Admins also get code indexing controls and AI privacy controls. Graphite is SOC 2 Type II compliant and continuously penetration tested, with GitHub tokens encrypted at rest under AES 256 and keys held in AWS Secrets Manager. Its AI features are opt in and do not train on customer code. The security page now sends SOC 2 report requests to trust.cursor.com. Sourcegraphite.com/pricing and graphite.com/docs/privacy-and-securityread 2026-10-28 |
||
|
Observability & Auditability Traces, logs, execution histories, metrics, audit events, and debugging detail for production agent behavior. |
||
|
cubicObservability & Auditability The Analytics API returns per pull request metrics (issues flagged and fixed, AI authored lines, engagement times), the Analytics AI review tab rebuilt on 6 Oct 2026 shows what happens to each issue cubic raises, Pro tracks AI coding usage, and Enterprise exports compliance audits. These report outcomes, and there is no stated per run record of the review agent's steps or tool calls. Sourcedocs.cubic.dev/analytics/api, docs.cubic.dev/changelog and cubic.dev/pricing-plansread 2026-10-28 |
||
|
GraphiteObservability & Auditability Team insights ship from Starter, with custom analytics on Enterprise, and give engineering leaders throughput and cycle time reporting. An audit log with SIEM export is available on Enterprise. A PR inbox centralizes reviews, review findings persist as durable inline comments on the pull request, and CI summaries explain failures in place. These report on team throughput and account access. Graphite names no trace of each step in the review agent's own reasoning, and no agent execution record, session history or replay. Sourcegraphite.com/pricing and docs/insightsread 2026-10-28 |
||
|
Deployment & Data Residency Deployment modes and options, including SaaS, dedicated cloud, VPC, on-prem, hybrid, local runtime, and self-hosting. |
||
|
cubicDeployment & Data Residency cubic is hosted SaaS installed as a GitHub app, and code is never stored or used for training. There is no stated self hosting, customer cloud or region choice. Sourcecubic.dev/pricing-plans, cubic.dev/enterprise and docs privacy-securityread 2026-09-29 |
||
|
GraphiteDeployment & Data Residency The product is hosted software on AWS, delivered through a GitHub app with a CLI and VS Code extension as clients. The Enterprise tier adds GitHub Enterprise Server support and private uploads, which let Graphite work with a code host the customer runs itself. Graphite itself does not run in the customer's environment. There is no self hosted, on premises, VPC or private instance option on any tier, Enterprise included, and customers cannot choose a region. Sourcegraphite.com/pricing and docs/privacy-and-securityread 2026-10-28 |
||
| Solution readiness | ||
|
Prebuilt Agents, Templates & Packs Ready-made workflows, packaged employees, templates, blueprints, industry solutions, and role-specific agents that reduce time-to-value. |
||
|
cubicPrebuilt Agents, Templates & Packs Teams turn on prebuilt review agents from cubic's agents library, which lists 63 agents, and adopt community and suggested agents with one click. Since 9 October the library runs on all repositories by default. Teams also write their own custom agents in natural language, five on Team and ten on Pro. Sourcecubic.dev/agents and docs.cubic.dev/ai-review/custom-agentsread 2026-10-25 |
||
|
GraphitePrebuilt Agents, Templates & Packs A single agent, Graphite Agent, does the reviewing. The custom rule dialog offers built in templates for common scenarios, such as language specific style guides, security best practices and accessibility standards, that a team picks and adapts. The template set is named by category only, with no full list. Sourcegraphite.com/docs/ai-review-customizationread 2026-10-28 |
||
| Platform extensibility | ||
|
Model Flexibility & Routing Ability to work across multiple foundation models, route tasks to different models, or let buyers bring their own providers and keys. |
||
|
cubicModel Flexibility & Routing Hosted pull request reviews run on models cubic chooses from OpenAI and Anthropic, so more than one provider is disclosed but the customer does not pick for the reviewer that acts on pull requests. Enterprise offers bring your own API keys without a model or provider selection. The cubic CLI alone lets a developer connect a ChatGPT or Claude Code subscription and switch providers and models for local reviews (changelog 19 Apr 2026), a picker that does not reach the hosted reviewer. Sourcedocs.cubic.dev/account/privacy-security, cubic.dev/pricing-plans and docs.cubic.dev/changelogread 2026-10-28 |
||
|
GraphiteModel Flexibility & Routing The AI features run on models from Anthropic and OpenAI, Graphite's two AI subprocessors, and the terms of service say either may generate outputs. Graphite makes the model choice itself and names no model picker, routing setting or bring your own key option for Graphite Agent; the model menu in the Agents tab belongs to Cursor Cloud Agents. Sourcegraphite.com/docs/ai-privacy-and-security, terms-of-service and docs/agentsread 2026-10-28 |
||
|
APIs, SDKs & MCP Extensibility Composability layer: stable APIs, SDKs, MCP tool consumption/serving, custom tools, and integration into internal systems. |
||
|
cubicAPIs, SDKs & MCP Extensibility cubic's own MCP server at cubic.dev/api/mcp, signed in with OAuth or an API key, offers 20 tools plus three environment tools, including get_pr_issues, trigger_pr_review, update_pr_issue_status, list_wikis, list_scans and get_scan. Several of them start reviews or change finding status, so outside agents can act through cubic, not only read from it. On Pro, an Analytics API adds a bearer token endpoint, and since 2 October organization API keys and a Members API are available. Sourcedocs.cubic.dev/ide/mcp-server and changelog 2026-10-25, analytics/apiread 2026-09-29 |
||
|
GraphiteAPIs, SDKs & MCP Extensibility GT MCP, in beta and built into the Graphite CLI from version 1.6.7, lets AI agents such as Cursor and Claude Code create stacked pull requests through Graphite's stacking operations, and MCP ships on every plan including the free Hobby tier. The CLI is a substantial client for stack management and simplified git commands, and Graphite publishes source on GitHub under the withgraphite organization. Graphite names no public REST API, SDK or webhooks for building on the platform. Sourcegraphite.com/docs/gt-mcp and graphite.com/pricingread 2026-10-28 |
||
|
Testing, Debugging & Optimization Testing, debugging, scoring, retries, fallbacks, quality gates, and optimization loops for improving agent workflows before and after deployment. |
||
|
cubicTesting, Debugging & Optimization Shadow mode keeps cubic in comment only mode on real pull requests while adding a summary of whether it would have approved each one, so a team tests the agent's approval behavior under its chosen policy before granting it live authority. Bug detection, the product itself, tests the customer's code and not the agent. Sourcedocs.cubic.dev/ai-review/auto-approvalread 2026-09-29 |
||
|
GraphiteTesting, Debugging & Optimization AI reviews check every pull request for bugs, logic errors and security issues before merge, with suggested fixes, CI summaries and rules the team tunes, and a dashboard tracks acceptance rates and rule effectiveness. The review works as a quality gate in the release path, where customers test their own changes. Sourcegraphite.com/docs/ai-reviews and graphite.com/pricingread 2026-10-28 |
||
| Specialist automation | ||
|
Browser & Computer Use Browser, desktop, or remote/local computer control for workflows that cannot be handled through stable APIs alone. |
||
|
cubicBrowser & Computer Use cubic works through the GitHub API, an isolated review container, a CLI and MCP, with no stated browser control or operation of software through a graphical interface. Sourcedocs.cubic.dev and cubic.dev/pricing-plansread 2026-09-29 |
||
|
GraphiteBrowser & Computer Use Every Graphite action runs through programmatic interfaces. These are the GitHub API for pull requests, comments, suggested fixes and merges, git for stacking and automatic rebase of stacks, the CLI, the VS Code extension, MCP, and CI integrations with Buildkite and GitHub Actions. Graphite names no browser control, screenshot capture or visual verification, and no way to operate software that lacks a programmatic interface. The Cursor Cloud Agents surfaced in the Agents tab run on Cursor's own machines and are Cursor's product. Sourcegraphite.com/pricing, graphite.com/features and docs/agentsread 2026-10-28 |
||
Pricing snapshot
Sourced from the Index pricing dataset · open each vendor's profile for full detail.
| Pricing | c cubic |
G Graphite |
|---|---|---|
|
Entry price Lowest public entry point |
Free plan. Team at $30 per developer per month billed annually, or $40 monthly. | Free (Hobby: stacking CLI, VS Code, limited AI reviews) · Starter $20/user/mo annual · Team $40/user/mo annual (unlimited Graphite Agent + merge queue) · Enterprise custom |
|
Pricing confidence How public the numbers are |
Public, exact | Public, exact |
|
Billing Primary billing axis |
Subscription tiers priced per developer per month, each with a monthly allowance of reviewed lines (the added and deleted diff lines cubic reads). The free tier is capped by monthly review count, and Enterprise is custom. | Per user per month subscription tiers billed annually, with a free tier and a 30 day trial. The Team tier includes unlimited AI reviews and chat with no per review metering; enterprise is custom. |
|
Variable cost Workload / overage exposure |
Medium variable cost | Low variable cost |
|
Free tier / trial Try before you buy |
Free tierTrial
|
Free tierTrial
|
|
Buying motion Self-serve vs sales call |
Self-serve | Self-serve |
|
Pricing verified When this pricing was last checked |
Product updates
Newest first, each linked to its source.
cubic
-
· Agent capability
Codebase scan agents in cubic can now check whether production data confirms the bugs and vulnerabilities they find, starting with PostHog, which is switched on per repository under Production evidence.
Source -
· Human approval / guardrails
cubic added a repository policy that allows automatic pull request approval while findings at selected priorities remain unresolved.
Source -
· Agent capability
cubic now supports cross-repo reviews, allowing users to link up to five related repositories for coordinated updates.
Source
Graphite joined Cursor after a December 2025 agreement with Anysphere, the maker of Cursor, and Anysphere became a SpaceXAI division in August 2026, so Graphite is not an independent vendor.
More comparisons with cubic or Graphite
Other matchups in coding agents
Not the pairing you were after? These compare a different set of coding agents on the same 14 capabilities.