Comparison

Fleet vs. Sourcegraph Amp: Agent Fleet Orchestration vs. AI Coding Agent with Codebase Context

Sourcegraph Amp is a hosted AI coding agent with deep codebase search and context retrieval. Fleet is a self-hosted orchestration layer for a team of specialized agents with approval gates, handoffs, and no dependency on Sourcegraph's infrastructure.

Sourcegraph Amp is a standalone AI coding agent from Sourcegraph, installed via its own CLI or a VS Code extension and billed pay-as-you-go — it does not require an existing Sourcegraph deployment. It draws on strong codebase context and code search to give agents accurate context before writing code — addressing one of the main failure modes of context-window-limited agents.

Fleet approaches context differently: agents receive focused prompts with role-specific context, and the watcher daemon ensures each agent is dispatched with the right task and environment variables. Fleet's strength is coordination and governance across a team of agents, not deep code search. The two tools can run side by side.

Choose Fleet if

Teams that want a self-hosted, role-based agent coordination layer with governance, budget controls, and reactive handoffs — independent of any commercial code intelligence platform.

Choose Sourcegraph Amp if

Developers who want a standalone AI coding agent with strong codebase context and code search, installed via CLI or VS Code on pay-as-you-go pricing.

Fleet vs. Sourcegraph Amp: side by side

Feature	Fleet	Sourcegraph Amp
Deployment	Self-hosted binary on your infrastructure	Standalone CLI + VS Code extension; runs against a hosted backend
Codebase context	Focused per-agent prompts and env vars	Deep code search and cross-repo context retrieval
Multi-agent support	Full role-based roster with handoffs	Single agent per session
Agent handoffs	Fabric event bus: dev -> reviewer -> release manager	Not supported
Model choice	Different Claude models per agent role (Opus/Sonnet/Haiku)	Amp-managed model selection
Data residency	Self-hosted; your source code goes only to your model backend (Bedrock/Vertex in your own cloud) + GitHub	Code processed by Amp's hosted backend
Governance	Budget caps, risk scoring, quarantine, audit log	Sourcegraph platform controls
Standalone install	Single Go binary	Standalone CLI + VS Code extension, no Sourcegraph deployment required

Where Fleet is the better fit

Self-hosted orchestration — Fleet runs entirely on your own infrastructure with no hosted backend in the loop
Multi-role pipeline automation with reactive handoffs that Amp's single-agent model does not provide
Self-hosted with no code transmitted to third-party infrastructure beyond model API calls
Per-role Claude model assignment (Opus/Sonnet/Haiku) plus run-time budgets let teams optimize cost across the pipeline

Where Sourcegraph Amp is the better fit

Cross-repo code search gives agents accurate context for large, multi-repo codebases — a genuine advantage over context-window-limited approaches
Strong codebase context and code search out of the box, with no orchestration platform to set up
Enterprise-grade code intelligence with precise references, hover documentation, and call graphs available to the agent
Strong option for organizations with strict code-graph requirements, e.g., large monorepos with complex dependency structures

Pricing

Sourcegraph Amp is pay-as-you-go on actual model usage, with a free tier — no Sourcegraph deployment required. Fleet offers a free tier (500 hosted starts) and Business at $299/org/month. Amp meters model usage; Fleet meters one hosted root-workflow start once, with model charges separate.

Do they compete, or coexist?

A Sourcegraph team can run Fleet agents that use the Sourcegraph search API (via src CLI) as part of their context-gathering step. Fleet handles coordination and governance; Sourcegraph provides the code intelligence layer. The integration is straightforward for teams running both.

Frequently asked questions

Does Fleet have code search?

Fleet does not include built-in code search. Agents receive task context via environment variables and focused prompts. For large monorepos where context retrieval is critical, integrating a code search tool like Sourcegraph or ripgrep into the agent's toolset is recommended.

Do I need Sourcegraph to use Fleet?

No. Fleet is fully independent of Sourcegraph and works with any git repository.

More comparisons

Fleet vs. Augment Code →Fleet vs. Cline →

Keep your AI agents from escaping

Jail Fleet-launched Linux agents, then run saved workflows with review, approvals, and an audit trail. Prove it at /security/#containment.

See how it works Install Fleet