Agent Atlas / Coding agents

Kilo-Org/kilocode

An open coding-agent platform spanning VS Code, JetBrains, a CLI, cloud execution, code review, multiple modes, MCP, and many model providers.

Coding agent platform Full dossier Project documented Reviewed July 21, 2026
Intended operator

Developers moving among IDE, terminal, cloud, and automated review surfaces.

Primary surfaces

VS Code and JetBrains / CLI and cloud agent / automated code review

Tracked repository

Kilo-Org/kilocode

Teaching mechanism

Autonomous mode deliberately removes confirmation prompts

The project documents an automation mode in which permission prompts are disabled, making the trust placed in the surrounding CI or sandbox environment explicit.

Evidence limitProject documentation states the contract; TheGitReporter has not yet traced the enforcement path through the implementation.
Inspect the source

System Map

Eight inspection boundaries
Model boundary

A broad provider catalog and mid-task switching separate the coding surface from one model vendor.

Context and memory

Project context, specialized modes, and custom agents shape the active task envelope.

Tools and execution

File editing, terminal, browser, and MCP capabilities cross IDE, CLI, and cloud surfaces.

Permissions and isolation

Interactive surfaces use prompts while documented autonomous mode delegates trust to the surrounding environment.

Sessions and recovery

Checkpoint and cross-surface continuity require deeper source inspection.

Delegation

Specialized and custom agents divide work by role; worker lineage remains an inspection question.

Human control

Plan, ask, debug, code, and review modes give the operator distinct work surfaces.

Evidence surface

README contracts, repository packages, security policy, tests, and future Daily reporting form the evidence trail.

Learn Through This Project

9 lessons

Compare Its Boundaries

8 labs

Latest Reporting

0 articles

No Daily Edition reporting is attached yet.

Primary Sources

Open Questions

  • Do IDE, CLI, cloud, and autonomous modes preserve one legible permission model?

Evidence state describes what this Atlas profile can support today. It is not a product score and it is not evidence that uninspected capabilities are absent.