Overview
Product statement
Syndicate Code is a governed AI development environment for operators who need fast coding workflows with strict control-plane enforcement. It is provider-agnostic, language-agnostic, and TUI-first, and it is not an unrestricted autonomy platform.
v1 deployment note This page describes the v1 local-embedded deployment. Team and enterprise deployments use
syndicate-serverand differ in control-plane topology, authentication, and remote endpoint operations.
Why governance matters
Most coding tools optimize for speed, but they do not expose a durable control-plane decision path for approvals, permits, checkpoints, and evidence. That gap makes incident response and post-hoc verification difficult.
Syndicate Code closes that gap with governed execution: proposals are canonicalized, evaluated by policy, checkpointed when required, bound to permits, and recorded in an append-only audit chain. This framing comes from product_system_definition.md §6 and governed_execution_spec.md §4.
The TUI is an operator surface, not an enforcement layer. The control plane remains authoritative, and runtime behavior is non-authoritative (product_system_definition.md §5; operator_interface_spec.md authority model).
The six-stage governed execution pipeline
- Proposal intake
- Canonicalization
- Policy evaluation
- Checkpoint (if required)
- Permit issuance
- Execution and evidence recording
Normative basis: governed_execution_spec.md §5 and syndicatecode-docs-site-spec.md §4.1.
What Syndicate Code guarantees
- G1 Governed execution: No side-effecting operation executes without control-plane authorization.
- G2 Binding: Approvals bind to canonicalized requests and cannot be reused outside scope.
- G3 Inspectability: Model context and prompts are reconstructable and inspectable.
- G4 Auditability: Decisions, authorizations, and executions are recorded as append-only events.
- G5 Provider abstraction: Provider selection is control-plane governed and not hardcoded in workflows.
- G6 Bounded autonomy with checkpoints: workflows remain bounded, attributable, interruptible, and checkpointed.
Normative basis: product_system_definition.md §6.
What Syndicate Code does not guarantee
Syndicate Code does not guarantee:
- correctness of model outputs
- absence of hallucinations
- prevention of all data exfiltration risks
- equivalence of audit fidelity across providers
Non-guarantee callout Governance controls authorization and evidence. They do not make model outputs inherently correct.
Normative basis: product_system_definition.md §7.
Competitive reference
| Axis | Cursor | Claude Code | Aider | Syndicate Code |
|---|---|---|---|---|
| Multi-file editing | Yes | Yes | Yes | Yes + governed execution |
| Any language | Yes | Yes | Yes | Yes + registry-driven |
| Any provider | No | No | Yes | Yes + models catalog |
| Streaming output | Yes | Yes | Yes | Yes + evidence stream |
| LSP integration | Yes | No | No | Yes |
| Audit trail | No | No | No | Yes (hash-chained HMAC evidence) |
| Checkpoint governance | No | No | No | Yes |
| Local inference | No | No | Yes | Yes |
| TUI-first | No | Yes | Yes | Yes (five-pane governed surface) |
| CI/headless mode | No | Yes | Yes | Yes |
| Team/enterprise deployment | No | No | No | Yes (syndicate-server) |
Source: syndicate_master_spec_and_plan.md Part I, Competitive Bar.
Quick navigation
| Section | Use this when |
|---|---|
| Getting Started | You want installation and first governed run guidance. |
| Using Syndicate Code | You need day-to-day workflow operation references. |
| Policy | You define trust tiers, capability rules, and checkpoint policy. |
| Audit | You verify chain integrity and evidence completeness. |