← All agent and harness briefs
Agent runtimeagenticcodingdocumentedunattended capable

Operational brief

Cline

Open-source coding harness across IDE, CLI and desktop, with optional scheduled work and hosted browser agents.

Official Cline image from cline.bot

Choose it when

You want an open coding agent in the editor, terminal, or desktop, model-agnostic.

Skip it when

You need local rules, skills, MCP and connectors to carry into hosted cloud sessions, or guaranteed human approval before every unattended write.

01 / Classification

Autonomy evidence

This is our classification, not a vendor badge. The record says what the product normally does. The gaps stay visible.

How we classify it
Tool use + multi-step plans, still human-gated per session.
Runs unattended
Yes — that capability is documented in this brief.
Trigger
Normal user-started IDE/CLI/desktop sessions; optional CLI/SDK/Kanban cron schedules run through the hub. Desktop separately documents timed/recurring Schedule tasks; this isn't an IDE schedule feature.
State
Hub schedules persist across process restarts and expose execution history. Cloud agents push their own branch and pause/resume idle sandboxes with workspace intact; that's separate from local schedule state.
Tools
File edits, shell commands and optional MCP tools. Scheduled hub runs approve their own calls in every mode, including plan; plan isn't a general no-write shell boundary.
Human control
Normal IDE use supports human approval and configurable auto-approve. For unattended work, the operator must inspect run history/last failures, set timeouts and pause schedules. Cloud work goes to a branch/PR for human review and merge; no automatic recovery guarantee is established.

02 / Operations

Control reality

Useful agents eventually touch something real. This is the operational picture we have, not a promise hidden in a sales page.

Where it runs
Local IDE/CLI/desktop and separately hosted browser sandboxes. Desktop docs describe macOS and Windows, with Windows in beta; its Schedule settings aren't the CLI's permission-default contract.
Data boundary
Local harness doesn't mean local inference: select a provider or supported local model. Cloud sessions omit local rules, skills, MCP servers and connectors; don't assume local controls follow a task to cloud.
Identity & permissions
Hosted tasks use a connected GitHub account and selected repository; Cline pushes a task branch, while you review and merge the PR. Provider/account configuration is surface-specific.
What it can write
repo files · terminal · MCP-connected systems where enabled
Surfaces
IDE extension · CLI · Desktop app · Hosted browser agents
Export / lock-in
Not recorded yet.

03 / Receipts

Evidence ledger

Here is what this brief is built on. “Documented” means we found a source. “Tested” means this desk actually ran it. Those are different things.

Evidence status
documented
Source set
8 primary sources
Evidence review
2026-10-14

Sources on file

Corrections

  • 2026-10-14 · Replaced June-only coverage with current surface-specific docs. Corrected unattended:false and hosted-only skip wording; added optional hub/Desktop scheduling and cloud boundaries without applying CLI defaults to Desktop. Latest API is Desktop v0.0.46, not an extension version; no fresh runtime test.

Desk notes

What we found

Useful context, not a substitute for the source record above. This is not a desk test unless the evidence status says tested.

Choose for model choice across local surfaces; scope unattended permissions yourself and review cloud PRs.

Primary mode
User-started agentic coding
Optional automation
Hub cron; separate Desktop Schedule
CLI approval
Auto-approve defaults true
Release scope
Desktop v0.0.46, not extension

What the vendor says

Scheduling docs distinguish CLI/SDK/Kanban from IDE; Desktop and hosted browser tasks have separate controls.

GitHub

Repository README describes a shared agent engine; latest release response is explicitly Desktop, not extension-wide currency.

Sources fetched 2026-10-14

04 / Change log

Related coverage

The brief tells you what the product is. Payloads track the changes that may alter the decision.