Skip to content
/

Dimensions

25 engineering dimensions

The RUBRIC-v3 schema evaluates each framework across 25 engineering dimensions. Each dimension page lists which frameworks have it (and which don't), sorted by stars. The numbers reveal which engineering rigor is common, which is rare, and which is unique.

TDD enforcement featured

Whether the framework enforces test-driven development — and how structurally: from prompt suggestion to compiled invariant.

185 / 653
Multi-agent

Frameworks spawning N concurrent agents (swarms, hierarchies, director/coder splits).

311 / 653
Multi-model routing

Per-role model selection — haiku for fast, opus for careful, cross-vendor for cost arbitrage.

220 / 653
Audit log

Persistent, queryable record of agent actions for compliance and replay.

294 / 653
Replay capability

Ability to re-run a past session deterministically from its audit log.

146 / 653
Crash recovery

Automatic recovery from agent loops, context exhaustion, or tool failures.

237 / 653
Worktree per feature

Each task runs in an isolated git worktree, enabling true parallel development.

83 / 653
Compaction handling

Explicit strategy for when context window fills — PreCompact hooks, summarization, handoff.

189 / 653
Cross-session handoff

State transfer between agent sessions, enabling long-horizon tasks.

413 / 653
CLI binary

Dedicated CLI binary (vs. just being a config file bundle).

308 / 653
MCP server

Bundles one or more MCP (Model Context Protocol) server tools.

165 / 653
Local UI

Ships a local web/desktop dashboard (not just a terminal interface).

247 / 653
Isolation mechanism

How the framework isolates tasks — git worktrees, Docker containers, sandboxes, PTYs.

248 / 653
Prompt chaining

Structured multi-turn interactions where outputs feed into subsequent prompts.

322 / 653
Spec format

How the framework encodes specifications — Markdown, YAML, XML, symbolic notation, DSL.

315 / 653
Auto-commits

Framework commits to git automatically after task completion.

98 / 653
Auto PR creation

Framework opens a pull request automatically.

71 / 653
Auto-merge on CI pass

Framework merges the PR automatically when CI passes.

20 / 653
Streaming output

Outputs stream in real-time rather than batching at completion.

212 / 653
Memory type

Persistence strategy: file-based, SQLite, vector, HTTP-proxy, episodic.

478 / 653
Cross-tool portability

How broadly the framework targets different AI agents (single-tool vs. multi-runtime).

586 / 653
Distribution type

How the framework is packaged — Claude plugin, npm, pip, git clone, YAML bundle.

618 / 653
Approval gates

Human-in-the-loop checkpoints before the agent proceeds to the next phase.

347 / 653
Workflow phases

Named phases of work (plan, spec, implement, review, verify) vs. a single-phase agent.

559 / 653
Orchestration pattern

How multiple agents relate — queen+workers, director/coder, peer-to-peer, pipeline.

394 / 653