Competitive Discovery — Human Review Required (Run #2, 2026-03-23)
The weekly discovery cycle evaluated 11 new candidates. 3 were auto-added to the watchlist (superset, mux, 1code — all 8/8). The remaining 8 require human review — 6 are high-confidence but hit the per-cycle auto-add cap of 3, and 2 are medium-confidence.
Auto-added this cycle ✅ (no action needed)
superset-sh/superset (score 8/8) — 7.7k stars, "IDE for the AI Agents Era", parallel agent orchestration
coder/mux (score 8/8) — 1.4k stars, 52 releases, by Coder Inc., multi-model agent multiplexer
21st-dev/1code (score 8/8) — 5.3k stars, MCP marketplace, cloud agents, REST API automation
Previously pending from Issue #470 (Run #1)
Issue #470 remains open with 8 candidates from the previous cycle. Those candidates are not duplicated here.
Candidates needing your decision
High-confidence (cap-limited, score ≥ 5):
| # |
Repo |
Stars |
Score |
One-line rationale |
| 1 |
HKUDS/ClawTeam |
2,800 |
7/8 |
Self-organizing agent swarms with worktrees + tmux + ZeroMQ P2P transport. Autonomous ML research use case. |
| 2 |
AutoMaker-Org/automaker |
3,100 |
8/8 |
Autonomous dev studio: Kanban → Claude Agent SDK → code → commit in worktrees. 2,278 commits. |
| 3 |
johannesjo/parallel-code |
396 |
7/8 |
Electron app for parallel Claude Code/Codex/Gemini in worktrees. v1.1.0, 638 commits, MIT. |
| 4 |
collaborator-ai/collab-public |
901 |
5/8 |
Infinite-canvas desktop app for agentic development. Terminals + files on a single canvas. |
| 5 |
phuryn/swarm-protocol |
31 |
5/8 |
MCP-native coordination protocol with intent/claim/signal primitives. Alpha, PostgreSQL-backed. |
| 6 |
ikamensh/kodo |
39 |
5/8 |
Autonomous overnight coding orchestrator (architect/worker/tester roles). Claims 57% SWE-bench. |
Medium-confidence (score 3-4):
| # |
Repo |
Stars |
Score |
One-line rationale |
| 7 |
farol-team/gnap |
33 |
3/8 |
Git-native agent protocol RFC — zero-server coordination via .gnap/ directory. Interesting architecture. |
| 8 |
sentientwave/automata |
44 |
3/8 |
Multi-agent via Matrix chat. Elixir. Not coding-specific but agent swarm model is adjacent. |
To act
Reply with numbers to approve (e.g., "1, 2, 3") or "none" to reject all. Approved entries will be added as source: "auto-discovered" in the next cycle.
Watchlist status after this cycle
- Total entries: 19
- User-authored: 6
- Auto-discovered: 4/10 (cap has room for 6 more)
- discovery-2026-03-15: 9 (legacy source label)
Generated by competitive-intel-analyst, Run #2 (2026-03-23)
🤖 Created by competitive-intel-analyst via Optimus Spartan Swarm
Competitive Discovery — Human Review Required (Run #2, 2026-03-23)
The weekly discovery cycle evaluated 11 new candidates. 3 were auto-added to the watchlist (superset, mux, 1code — all 8/8). The remaining 8 require human review — 6 are high-confidence but hit the per-cycle auto-add cap of 3, and 2 are medium-confidence.
Auto-added this cycle ✅ (no action needed)
superset-sh/superset(score 8/8) — 7.7k stars, "IDE for the AI Agents Era", parallel agent orchestrationcoder/mux(score 8/8) — 1.4k stars, 52 releases, by Coder Inc., multi-model agent multiplexer21st-dev/1code(score 8/8) — 5.3k stars, MCP marketplace, cloud agents, REST API automationPreviously pending from Issue #470 (Run #1)
Issue #470 remains open with 8 candidates from the previous cycle. Those candidates are not duplicated here.
Candidates needing your decision
High-confidence (cap-limited, score ≥ 5):
HKUDS/ClawTeamAutoMaker-Org/automakerjohannesjo/parallel-codecollaborator-ai/collab-publicphuryn/swarm-protocolikamensh/kodoMedium-confidence (score 3-4):
farol-team/gnapsentientwave/automataTo act
Reply with numbers to approve (e.g., "1, 2, 3") or "none" to reject all. Approved entries will be added as
source: "auto-discovered"in the next cycle.Watchlist status after this cycle
Generated by competitive-intel-analyst, Run #2 (2026-03-23)
🤖 Created by
competitive-intel-analystvia Optimus Spartan Swarm