The persistent, model-agnostic harness
for coding agents.

What is a coding agent harness?

A coding agent harness is the software around a model that supplies context, manages tools and execution, tracks task state, and defines how proposed changes are verified and accepted. The model generates the code; the harness establishes the conditions under which that code can enter the project.

Snipara makes that workflow context-aware. Repository sources, reviewed decisions and project memory travel into the task, while proof and review receipts travel back with the candidate.

The agent can change. The work continues.

Claude Code created a plan and completed the first two phases. A fresh Codex session resumed the saved Companion checkpoint and finished the next phase. The plan, decisions, completed tasks, attempt counters and original evidence stayed intact.

21 independent validation cases passed on October 10, 2026. One local fixture, using Companion 3.7.0 and a shared workspace with operator-assisted command dispatch.

Resume requires access to the project folder, including saved .snipara state and workflow artifacts, and the project’s Snipara MCP. This demonstration does not qualify live-process migration, MCP-only restoration or transfer of an Orchestrator controller between coding clients.

Independent checks ran outside the producing Codex session. This is an operator-reported demonstration, not a reliability benchmark or authenticated human acceptance.

One paired continuation, with and without Snipara

Both fresh Codex sessions received the same Claude-produced source, plan, decisions and ASCII evidence. The Snipara session also received the saved checkpoint and Hosted MCP access. Both passed; this fixture showed no speed or token saving.

The Snipara session includes one network-permission retry. Successful command timings exclude the failed probe; full session time includes it. Session time includes implementation and is not resume latency. Avoided task rework, time to first useful progress and total harness overhead were not measured. One small fixture is insufficient for a general performance claim.

Reproduce the scenario: scripts, frozen inputs and both candidates · Download the paired result and token counts

From project knowledge to a checked change.

Define the allowed files, required proof and execution limits before dispatch. The native host runs the agent with its own permission model. External validation checks the result; a separate reviewer can refuse it even when tests pass. An integration candidate stays bound to the reviewed tree and patch.

Bounded proof retries are opt-in. Controlled integration requires explicit policy and acceptance gates; stopping for human attention is a valid outcome.

Three layers, one workflow.

Choose the integration your task needs.

These are configuration-specific observations, not a universal model compatibility or reliability guarantee. Read the validation scope.

When to use the full harness.

Use the context layer for a quick lookup or a small edit. Add managed execution when a coding task spans phases, needs a separate reviewer, or must produce verifiable evidence before integration. Your model and coding client keep their reasoning and tools; Snipara provides project knowledge and workflow controls.