An opinionated atlas of frameworks
Choosing between Superpowers, SpecKit, OpenSpec, BMAD, and Taskmaster?
Start here. The atlas reads 654 entries — 348 frameworks built on top of Claude Code, Cursor, Codex, and others, plus 305 supporting tools they sit on, against a 25-dimension rubric — and explains, in plain language, what each one actually does differently.
§ focus
The five most-asked-about.
Each ships installable methodology — commands, skills, hooks, MCP, personas. Each card reads best when and skip if from its METRICS, so the design choice is up front, not in the docs.
Iron-law SKILL.md + SessionStart hook. The behavioral floor everyone else builds against.
GitHub's official toolkit. Spec → Plan → Tasks pipeline with 9 commands + 18 before/after hooks.
Spec-as-source-of-truth across 29 AI tools. Delta-diff specs, mirrored cmd+skill model.
Six named persona subagents — analyst, architect, dev, pm, sm, ux. Genuine multi-agent 'Party Mode'.
PRD → tasks.json → autonomous expansion. The analyze-complexity stage writes the next stage's prompts.
§ collisions
Why the field feels confusing.
Same name, wildly different things. The corpus has six independent projects called Conductor, eight or more oh-my-* repos with opposite purposes, and a long museum of similar collisions.
Six independent frameworks called 'Conductor' or 'claude-conductor' — a CC plugin (×5 variants), a YAML workflow engine, a macOS app, a GitHub-native swarm orchestrator.
Eight or more 'oh-my-claude / oh-my-codex / oh-my-claudecode' repos, split between orchestration frameworks and status-line customizers. Same prefix, opposite purposes.
Two implementations of the same Ralph loop pattern — snarktank's 200-line bash (19.6k★, dormant) and frankbria's 566-test production fork (9.2k★, active).
Two unrelated projects called 'Blueprint': Imbue's planning copilot and JuliusBrussee's cavekit framework with 75% token reduction via caveman notation.
The Stack · synthesis baseline
Five frameworks, one defensible reference design
If forced to pick five frameworks to merge into a flawless reference design — this is the Phase E answer. Each layer solves exactly one dimension; the five together cover all 25 rubric dimensions. Phase D adds two more for sub-agent context injection and security governance.
Source: _index/PHASE-E-EXECUTIVE-SUMMARY.md §5
This is the same 5-framework baseline from Phase C — Phase D did not surface a framework that replaces any on its own dimension. The defensible 7-framework set adds Trellis (LLM-evaluated hooks at the sub-agent boundary) and gh-aw-githubnext (security/governance reference). entroly is a memory tool (category: other) but is the only substrate-level memory framework in the corpus — included here with an explicit label.
§ entrances
Three ways in.
Different readers want different entry points. The Console for sorting; the archetypes for grouping; the tour for narrative.
A guided walk through the corpus — the five seed frameworks, the operational layer, the Phase D primitives, the synthesis baseline.
Sortable, filterable index of all 348 frameworks. Q-score column, archetype facets, side-by-side compare.
Grouped by use-case: methodology, multi-agent, memory, governance, substrate, specialized. The spine of the taxonomy.
§ spotlight
Twelve worth a deep look.
One per archetype family — curated for architectural distinctness, not stars. Stars are an SEO signal in this corpus; these twelve are the ones where the engineering depth matches.
Behavioral floor — iron-law SKILL.md + SessionStart hook.
Spec lifecycle — delta-diff specs mirrored across 29 AI tools.
Hive-mind queen + workers + 305 MCP tools + SQLite memory.
PR-lifecycle daemon — tracker poll + tmux PTY + CI reaction.
Cross-vendor router — Claude + Codex + Gemini with break-loop protocol.
Memory substrate — HTTP-proxy intercept with hallucination certificates.
JIT context — semantic on-demand fragment loading via MCP.
GitHub Next's safe-outputs separation + SHA-pinned deps.
LLM-evaluated hooks — Bug Council 5-analyst escalation.
Living spec — runtime failures auto-backpropagate as new criteria.
Sub-agent context injection — PreToolUse-on-Task hook.
Token-efficient encoding — PreToolUse Bash compression. 54.6k★.
Show top-by-stars instead (cautionary)
Here's what the same shortlist looks like when sorted only by GitHub stars. Notice the namespace-squatters dominate the top — that's why we curate.