Starkslab
Agent Tool RadarLive database runagent-tool-radar-v1.1

public GitHub evidence / operator scoring

Repositories worth an agent-tool operator look.

A production watch surface for agent-tool repositories with enough public momentum, fit, and activity to deserve an operator review.

Review candidatesSee scoring logic
Candidates
23
Top score
72
Repos
23
Snapshots
40
Endpoint runs
208
Docs
81
Live database runagent-tool-radar-v1.1

Serving scored candidates from Postgres run radar-20260522174245-84fc3450.

Latest completed radar run radar-20260522174245-84fc3450.

Generated May 22, 2026/Latest run May 22, 2026/Latest endpoint May 22, 2026/Latest snapshot May 22, 2026/Postgres radar tables from public GitHub data
Query matrix
9/9 enabled/208 endpoint runs/10 query rows/81 docs
Coverage lanes
agent runtimesCLI/SDKcoding agentscomputer-use/browser-useMCP toolsmemoryobservability/tracing/evalsorchestration/workflowssandbox/runtime

ranked candidates

Agent-tool repositories

Each row shows what the project is, why it surfaced, score components, evidence boundary, and the recommended next operator action.

#1RisingAgent runtimesPython
openai/openai-agents-python

A lightweight, powerful framework for multi-agent workflows

Stars
26.6K
Forks
4.1K
PRs
64
agentsaiframeworkllmpython
Release May 19, 2026/Snapshot May 22, 2026/8 docs/177 star samples/6 scored signals
Radar score
72
Rising
Source
69+20.6
Momentum
75+18.7
Activity
84+12.6
Fit
58+8.7
Novelty
65+6.5
Conversion
96+4.8
Why it surfaced

openai/openai-agents-python surfaced through Agent runtimes with strongest evidence in source signal and repository momentum.

Next action

Track in the next refresh cycle and compare component movement.

Evidence boundary

Evidence is limited to public GitHub metadata, 5 stored snapshots, 0 bounded query matches, 8 bounded document excerpts, 177 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#2RisingAgent runtime toolsCoding agentsTypeScript
callstackincubator/agent-device

CLI to control iOS and Android devices for AI agents

Stars
2.1K
Forks
122
PRs
1
adbagentic-aiagentsandroid-emulatorautomation
Release May 13, 2026/Snapshot May 18, 2026/9 docs/203 star samples/6 scored signals
Radar score
71
Rising
Source
76+22.8
Momentum
48+12
Activity
46+6.8
Fit
100+15
Novelty
90+9
Conversion
100+5
Why it surfaced

callstackincubator/agent-device surfaced through Agent runtime tools and Coding agents with strongest evidence in source signal and agent-tool fit.

Next action

Evaluate API, CLI, or MCP surface for a practical Starkslab angle.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 0 bounded query matches, 9 bounded document excerpts, 203 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#3RisingAgent runtime toolsCoding agentsTypeScript
openclaw/openclaw

Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞

Stars
372.9K
Forks
77.3K
PRs
3,501
aiassistantcrustaceanmoltyopenclaw
Release May 18, 2026/Snapshot May 18, 2026/8 docs/0 star samples/6 scored signals
Radar score
70
Rising
Source
76+22.8
Momentum
48+12
Activity
82+12.3
Fit
68+10.1
Novelty
79+7.9
Conversion
96+4.8
Why it surfaced

openclaw/openclaw surfaced through Agent runtime tools and Coding agents with strongest evidence in source signal and activity.

Next action

Evaluate API, CLI, or MCP surface for a practical Starkslab angle.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 0 bounded query matches, 8 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#4RisingCoding agentsTypeScript
anomalyco/opencode

The open source coding agent.

Stars
164K
Forks
19.4K
PRs
948
Release May 21, 2026/Snapshot May 22, 2026/8 docs/0 star samples/6 scored signals
Radar score
70
Rising
Source
71+21.2
Momentum
62+15.5
Activity
90+13.5
Fit
58+8.6
Novelty
65+6.5
Conversion
96+4.8
Why it surfaced

anomalyco/opencode surfaced through Coding agents with strongest evidence in source signal and repository momentum.

Next action

Track in the next refresh cycle and compare component movement.

Evidence boundary

Evidence is limited to public GitHub metadata, 5 stored snapshots, 0 bounded query matches, 8 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#5RisingCoding agentsRust
openai/codex

Lightweight coding agent that runs in your terminal

Stars
84.7K
Forks
12.3K
PRs
302
Release May 21, 2026/Snapshot May 22, 2026/8 docs/0 star samples/6 scored signals
Radar score
70
Rising
Source
71+21.2
Momentum
62+15.6
Activity
90+13.5
Fit
62+9.3
Novelty
65+6.5
Conversion
82+4.1
Why it surfaced

openai/codex surfaced through Coding agents with strongest evidence in source signal and repository momentum.

Next action

Evaluate API, CLI, or MCP surface for a practical Starkslab angle.

Evidence boundary

Evidence is limited to public GitHub metadata, 5 stored snapshots, 0 bounded query matches, 8 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#6RisingAgent runtimesPython
Arize-ai/phoenix

AI Observability & Evaluation

Stars
9.7K
Forks
879
PRs
54
llmopsai-monitoringai-observabilityllm-evalaiengineering
Release May 19, 2026/Snapshot May 19, 2026/8 docs/195 star samples/6 scored signals
Radar score
68
Rising
Source
70+21.1
Momentum
48+12
Activity
89+13.4
Fit
76+11.4
Novelty
52+5.2
Conversion
96+4.8
Why it surfaced

Arize-ai/phoenix surfaced through Agent runtimes with strongest evidence in source signal and activity.

Next action

Evaluate API, CLI, or MCP surface for a practical Starkslab angle.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 1 bounded query match, 8 bounded document excerpts, 195 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#7RisingCoding agentsTypeScript
n8n-io/n8n

Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.

Stars
188.5K
Forks
57.8K
PRs
1,050
aiapisautomationclidata-flow
Release May 18, 2026/Snapshot May 18, 2026/8 docs/0 star samples/6 scored signals
Radar score
67
Rising
Source
70+21.1
Momentum
48+12
Activity
81+12.2
Fit
84+12.6
Novelty
40+4
Conversion
96+4.8
Why it surfaced

n8n-io/n8n surfaced through Coding agents with strongest evidence in source signal and agent-tool fit.

Next action

Evaluate API, CLI, or MCP surface for a practical Starkslab angle.

Evidence boundary

Evidence is limited to public GitHub metadata, 2 stored snapshots, 0 bounded query matches, 8 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#8RisingAgent runtime toolsCoding agentsRust
t8y2/dbx

15MB, lightweight, cross-platform database client. Supports MySQL, PostgreSQL, SQLite, Redis, MongoDB, DuckDB, ClickHouse, SQL Server and more.15MB，轻量级跨平台数据库客户端。支持 MySQL、PostgreSQL、SQLite、Redis、MongoDB、DuckDB、ClickHouse、SQL Server 等。

Stars
1.8K
Forks
127
PRs
0
clickhousedatabasedatabase-clientdatabase-managementduckdb
Release May 22, 2026/Snapshot May 22, 2026/0 docs/0 star samples/6 scored signals
Radar score
65
Rising
Source
76+22.8
Momentum
46+11.5
Activity
58+8.7
Fit
56+8.4
Novelty
90+9
Conversion
96+4.8
Why it surfaced

t8y2/dbx surfaced through Agent runtime tools and Coding agents with strongest evidence in source signal and repository momentum.

Next action

Track in the next refresh cycle and compare component movement.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 0 bounded query matches, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#9RisingCoding agentsRust
Mesh-LLM/mesh-llm

Distributed AI/LLM for the people. Share compute privately or publicly to power your agents and chat.

Stars
1.1K
Forks
131
PRs
20
agentsaidecentralizeddistributedllm
Release May 22, 2026/Snapshot May 22, 2026/0 docs/0 star samples/6 scored signals
Radar score
65
Rising
Source
74+22.3
Momentum
39+9.6
Activity
78+11.7
Fit
55+8.3
Novelty
90+9
Conversion
84+4.2
Why it surfaced

Mesh-LLM/mesh-llm surfaced through Coding agents with strongest evidence in source signal and activity.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 1 bounded query match, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#10Low FitAgent runtimesRust
ruvnet/RuVector

RuVector is a High Performance, Real-Time, Self-Learning Ai, Vector GNN, Memory DB built in Rust.

Stars
4.1K
Forks
519
PRs
87
aiai-ocrgnngraphllm-inference
Release May 7, 2026/Snapshot May 22, 2026/0 docs/0 star samples/6 scored signals
Radar score
63
Low Fit
Source
70+21.1
Momentum
48+12
Activity
90+13.4
Fit
31+4.6
Novelty
79+7.9
Conversion
84+4.2
Why it surfaced

ruvnet/RuVector surfaced through Agent runtimes with strongest evidence in source signal and activity.

Next action

Keep as background signal; do not prioritize operator coverage yet.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 2 bounded query matches, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#11RisingCoding agentsPython
UrsushoribilisMusic/agentic-fleet-hub

Self-hosted orchestration layer for autonomous AI agent teams. Shared memory, heartbeat scheduling, vault-first secrets, and cross-model peer review — one command to deploy.

Stars
66
Forks
4
PRs
0
agent-orchestrationai-agentsai-infrastructureautonomous-agentsclaude-code
Release n/a/Snapshot May 18, 2026/8 docs/66 star samples/6 scored signals
Radar score
63
Rising
Source
75+22.5
Momentum
25+6.3
Activity
64+9.6
Fit
80+12
Novelty
90+9
Conversion
74+3.7
Why it surfaced

UrsushoribilisMusic/agentic-fleet-hub surfaced through Coding agents with strongest evidence in source signal and agent-tool fit.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 0 bounded query matches, 8 bounded document excerpts, 66 starred_at samples, and 30 recent public events. No LLM ranking or private usage data is used.

#12RisingCoding agentsGo
faramesh/faramesh-core

Governance-as-Code for AI agents. Open source policy layer with deterministic enforcement. Provisioning Identity, Brokering Credentials & Ensuring safe deployment

Stars
54
Forks
13
PRs
7
agentic-aiai-agentsdeterministicmcpagent-governance
Release Apr 15, 2026/Snapshot May 19, 2026/0 docs/0 star samples/6 scored signals
Radar score
60
Rising
Source
74+22.3
Momentum
21+5.3
Activity
63+9.5
Fit
70+10.5
Novelty
79+7.9
Conversion
96+4.8
Why it surfaced

faramesh/faramesh-core surfaced through Coding agents with strongest evidence in source signal and agent-tool fit.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 1 bounded query match, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#13RisingAgent runtimesTypeScript
humanlayer/humanlayer

The best way to get AI coding agents to solve hard problems in complex codebases.

Stars
10.9K
Forks
919
PRs
37
agentsaihuman-in-the-loopllmllms
Release Feb 27, 2026/Snapshot May 22, 2026/8 docs/0 star samples/6 scored signals
Radar score
59
Rising
Source
69+20.6
Momentum
32+8.1
Activity
60+9
Fit
70+10.5
Novelty
57+5.7
Conversion
96+4.8
Why it surfaced

humanlayer/humanlayer surfaced through Agent runtimes with strongest evidence in source signal and agent-tool fit.

Next action

Evaluate API, CLI, or MCP surface for a practical Starkslab angle.

Evidence boundary

Evidence is limited to public GitHub metadata, 5 stored snapshots, 0 bounded query matches, 8 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#14RisingAgent runtimesTypeScript
joshuaswarren/remnic

Open-source memory and context for user-aware agents: scoped memory, provenance, retrieval quality, correction, boundaries, evals, and MCP/HTTP access.

Stars
73
Forks
11
PRs
2
ai-agentai-memoryconversational-aiengramknowledge-graph
Release May 19, 2026/Snapshot May 19, 2026/0 docs/0 star samples/6 scored signals
Radar score
59
Rising
Source
72+21.7
Momentum
22+5.4
Activity
58+8.7
Fit
66+9.8
Novelty
90+9
Conversion
82+4.1
Why it surfaced

joshuaswarren/remnic surfaced through Agent runtimes with strongest evidence in source signal and agent-tool fit.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 2 bounded query matches, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#15RisingCoding agentsTypeScript
sverklo/sverklo

Local-first code intelligence — MCP server for Claude Code, Cursor, Windsurf, Zed. Symbol graph, blast-radius, git-pinned memory. 43× fewer tokens than naive grep. MIT, zero-config.

Stars
62
Forks
8
PRs
0
bm25claude-codecode-intelligencecode-searchcursor
Release May 22, 2026/Snapshot May 22, 2026/0 docs/0 star samples/6 scored signals
Radar score
59
Rising
Source
72+21.7
Momentum
21+5.3
Activity
55+8.2
Fit
66+10
Novelty
90+9
Conversion
96+4.8
Why it surfaced

sverklo/sverklo surfaced through Coding agents with strongest evidence in source signal and agent-tool fit.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 1 bounded query match, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#16RisingAgent runtime toolsCoding agentsTypeScript
MuhammadUsmanGM/code-stick

Plug in a USB. Get an offline AI coding agent on any laptop — Windows, macOS, Linux. opencode + Ollama, pre-built for 5 targets, zero install on the host.

Stars
49
Forks
9
PRs
0
coding-agentlocal-llmoffline-aiollamaopencode
Release n/a/Snapshot May 18, 2026/0 docs/0 star samples/6 scored signals
Radar score
59
Rising
Source
82+24.6
Momentum
21+5.3
Activity
36+5.3
Fit
76+11.5
Novelty
92+9.2
Conversion
70+3.5
Why it surfaced

MuhammadUsmanGM/code-stick surfaced through Agent runtime tools and Coding agents with strongest evidence in source signal and agent-tool fit.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 0 bounded query matches, 0 bounded document excerpts, 0 starred_at samples, and 17 recent public events. No LLM ranking or private usage data is used.

#17RisingCoding agentsTypeScript
openMF/web-app

Mifos X Web App is the Web-UI component of Mifos X. It currently uses Apache Fineract as the backend and leverages the popular Angular framework. Main Branch is the latest Stable Release, Dev should be used for all development ahead of release.

Stars
362
Forks
930
PRs
29
mifosx-platformangular-7angular-material-7typescriptscss
Release Dec 23, 2025/Snapshot May 19, 2026/0 docs/0 star samples/6 scored signals
Radar score
58
Rising
Source
72+21.7
Momentum
44+10.9
Activity
80+11.9
Fit
35+5.3
Novelty
42+4.2
Conversion
84+4.2
Why it surfaced

openMF/web-app surfaced through Coding agents with strongest evidence in source signal and activity.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 1 bounded query match, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#18Low FitAgent runtimesTypeScript
alchemy-run/alchemy-effect

Infrastructure as Effects

Stars
479
Forks
53
PRs
46
Release May 21, 2026/Snapshot May 22, 2026/0 docs/0 star samples/6 scored signals
Radar score
57
Low Fit
Source
72+21.7
Momentum
29+7.2
Activity
87+13
Fit
26+3.9
Novelty
79+7.9
Conversion
62+3.1
Why it surfaced

alchemy-run/alchemy-effect surfaced through Agent runtimes with strongest evidence in source signal and activity.

Next action

Keep as background signal; do not prioritize operator coverage yet.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 1 bounded query match, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#19WatchAgent runtime toolsCoding agentsRust
Dicklesworthstone/frankenlibc

Rust interposition layer for glibc: transparent safety membrane that incrementally replaces C library functions with memory-safe Rust implementations

Stars
43
Forks
3
PRs
0
libclinuxmemory-safetyrustsystems-programming
Release n/a/Snapshot May 18, 2026/0 docs/0 star samples/6 scored signals
Radar score
56
Watch
Source
80+24.1
Momentum
21+5.2
Activity
35+5.3
Fit
56+8.4
Novelty
92+9.2
Conversion
70+3.5
Why it surfaced

Dicklesworthstone/frankenlibc surfaced through Agent runtime tools and Coding agents with strongest evidence in source signal and novelty.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 0 bounded query matches, 0 bounded document excerpts, 0 starred_at samples, and 8 recent public events. No LLM ranking or private usage data is used.

#20WatchCoding agentsRust
c4pt0r/pie

Rust port of the pi agent harness — coding agent + LLM runtime stack

Stars
47
Forks
3
PRs
0
Release n/a/Snapshot May 22, 2026/0 docs/0 star samples/6 scored signals
Radar score
55
Watch
Source
70+21.1
Momentum
21+5.2
Activity
50+7.4
Fit
60+9
Novelty
88+8.8
Conversion
70+3.5
Why it surfaced

c4pt0r/pie surfaced through Coding agents with strongest evidence in source signal and agent-tool fit.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 0 bounded query matches, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#21WatchCoding agentsKotlin
livekit/client-sdk-android

LiveKit SDK for Android

Stars
337
Forks
140
PRs
9
androidkotlinvideowebrtc
Release May 5, 2026/Snapshot May 18, 2026/0 docs/0 star samples/6 scored signals
Radar score
51
Watch
Source
66+19.9
Momentum
30+7.5
Activity
66+9.8
Fit
35+5.3
Novelty
40+4
Conversion
80+4
Why it surfaced

livekit/client-sdk-android surfaced through Coding agents with strongest evidence in source signal and activity.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 0 bounded query matches, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#22Low FitCoding agentsRust
Trapezohe/Anamnesis

No repository description available.

Stars
90
Forks
14
PRs
0
Release May 18, 2026/Snapshot May 22, 2026/0 docs/0 star samples/6 scored signals
Radar score
49
Low Fit
Source
68+20.5
Momentum
22+5.5
Activity
41+6.2
Fit
30+4.5
Novelty
88+8.8
Conversion
70+3.5
Why it surfaced

Trapezohe/Anamnesis surfaced through Coding agents with strongest evidence in source signal and novelty.

Next action

Keep as background signal; do not prioritize operator coverage yet.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 0 bounded query matches, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

#23WatchAgent runtime toolsElixir
agentjido/ash_jido

Ash Framework integration with Jido Actions

Stars
38
Forks
10
PRs
0
Release n/a/Snapshot May 22, 2026/0 docs/0 star samples/6 scored signals
Radar score
47
Watch
Source
68+20.5
Momentum
21+5.3
Activity
34+5.1
Fit
41+6.2
Novelty
76+7.6
Conversion
54+2.7
Why it surfaced

agentjido/ash_jido surfaced through Agent runtime tools with strongest evidence in source signal and novelty.

Next action

Re-score after the next stored snapshot before making a conversion call.

Evidence boundary

Evidence is limited to public GitHub metadata, 1 stored snapshot, 0 bounded query matches, 0 bounded document excerpts, 0 starred_at samples, and 0 recent public events. No LLM ranking or private usage data is used.

scoring logic

Deterministic public signals

The Radar ranks public GitHub candidates from stored snapshots, source matches, and activity signals. JSON and source links stay available for verification, but they are diagnostics rather than the main journey.

Current limitations
  • Scores use public GitHub metadata and stored snapshots only.
  • Search is bounded by a typed query matrix; broad GitHub result sets are treated as top-rank evidence, not complete depth.
  • Stargazer timestamp samples are bounded to candidate/top-N repos and are not complete star history.
  • README/package/docs extraction stores bounded excerpts and skips oversized files; no broad cloning is used.
  • No LLM ranking, private analytics, or private GitHub traffic data is used.
Refresh cadence
  • Endpoint runs and rate buckets: recorded on every public GitHub request.
  • Search query matrix: 2-4 times per day for hot lanes, daily for secondary lanes.
  • Repository snapshots and GraphQL enrichment: daily, with hot repos every 4-6 hours.
  • README/package/docs extraction: daily for candidates and top-N scored repos.
  • Timestamped stargazer sampling: nightly for candidate/top-N repos only.
  • Public JSON/cache export: daily or twice daily.
Developer diagnostics

Raw data and implementation provenance are exposed for review, but they are not the primary reader path.

JSON feedSource repo