Developers / OSS

93 ideas, independently ranked for this context · top score 8.2 · 24 clear the 8.0 bar.

93 ideas

8.2

SandboxKit

Drop-in self-hosted agent sandbox plus auto-generated audit evidence so EU regulated teams run Claude agents where PII never leaves their VPC.

AI Security & GovernanceDevelopers

wow 7pas 9.05w

8.2

costwatch

A self-hostable LLM metering layer that tags every model call by feature/customer, alerts before a bill explodes, and exports usage records so you can bill your own users — any provider, drop-in proxy.

AI InfrastructureDevelopers

wow 8pas 8.54w

8.2

fattura-mcp

An MCP server plus TypeScript SDK that lets any AI agent issue, validate, and track Italian SDI e-invoices from natural language — generating compliant XML locally and routing transmission through an accredited partner channel.

AI Dev ToolsDevelopers

wow 8pas 8.56w

8.2

AgentReplay

Record a real agent session, freeze the tool responses into a deterministic fixture, and replay it in CI for free — with a hosted dashboard that flags flaky tests and fixture drift.

AI Dev ToolsDevelopers

wow 8pas 8.05w

8.2

ConsentLayer

Drop-in EU AI Act transparency for any generative app: C2PA provenance signing, watermarking, AI-disclosure UI, and an always-current compliance audit log.

AI Security & GovernanceDevelopers

wow 7pas 8.54w

8.2

PromptRegression

Git-style version control, diff, and regression testing for prompts and agent system messages — a one-word change runs your saved eval set in CI and shows before/after in the PR before it ships.

AI Dev ToolsDevelopers

wow 8pas 8.05w

8.2

accessibility-fixer-agent

A GitHub agent that gates every PR for EAA/WCAG regressions and opens reviewable fix PRs — with the legal obligation cited — so your app stays compliant by default.

AI Dev ToolsDevelopers

wow 8pas 8.05w

8.1

AgentStatusPage

The public trust layer for AI products — uptime, model availability, live eval pass-rate, and incident comms for the 'is it down or just being weird?' moment.

AI InfrastructureDevelopers

wow 8pas 7.53w

8.1

rag-eval-bench-eu

An open multilingual RAG eval benchmark with a CI gate that blocks merges degrading retrieval quality in European languages (IT/PT/ES/DE/FR).

Data & RAGDevelopers

wow 9pas 7.05w

#10

8.1

ClawShield

A supply-chain security scanner for AI agent skills and MCP servers (OpenClaw/ClawHub and beyond) that flags credential-leaking and malicious packages before you install — with a public trust registry the community references.

AI Security & GovernanceDevelopers

wow 8pas 7.03w

#11

8.1

SchemaGuard

A schema firewall for AI agents: validates, auto-repairs, and alerts on every LLM JSON output so a drifted field can never silently 500 your app.

AI Security & GovernanceDevelopers

wow 8pas 7.04w

#12

8.1

WellKnownGen

The agent-readiness platform for the web: a free one-click audit that emits every well-known file (llms.txt, WebMCP, LLMFeed, agent.json), plus paid monitoring that shows which AI agents actually crawl and buy on your site.

AI Dev ToolsDevelopers

wow 8pas 7.04w

#13

8.1

TrendRadar

A daily agent that watches GitHub trending, HN, and model release notes and ships vetted 'what just became buildable' product-gap briefs with a capability-unlock score.

Agents & AutomationDevelopers

wow 9pas 6.53w

#14

8.0

InboxAgent OSS

The hosted, safe, one-click way to run an OpenClaw-style personal agent gateway (WhatsApp/Telegram/email/calendar) — managed by default, with security guardrails so non-technical users never expose their accounts.

AI Security & GovernanceDevelopers

wow 8pas 8.05w

#15

8.0

DocStructure

An OSS pipeline that turns messy enterprise PDFs/scans/tables into clean, typed, agent-ready JSON — you bring your own vision-model keys, and it self-corrects against your schema with confidence scores and human-review flags.

AI Dev ToolsDevelopers

wow 8pas 8.06w

#16

8.0

CostLens

An AI FinOps agent that attributes your multi-provider LLM spend per feature and customer, then automatically routes traffic to cheaper models to cut the bill.

AI InfrastructureDevelopers

wow 7pas 8.55w

#17

8.0

openapi-to-mcp

Paste an OpenAPI spec, get a typed MCP server we host and keep continuously in sync with your spec — auth, rate limits, and pagination handled.

AI InfrastructureDevelopers

wow 8pas 8.05w

#18

8.0

PrivateMCP

Give agents secure, audited, read-only-first access to internal databases and tools over MCP Tunnels — no public ports, full policy and audit control.

AI Security & GovernanceDevelopers

wow 8pas 8.04w

#19

8.0

ComplianceCopilot

An EU AI Act copilot that drafts your technical documentation, DPIA, and conformity assessment, then keeps them current as the rules and your system change.

AI Security & GovernanceDevelopers

wow 7pas 8.55w

#20

8.0

forfettario-copilot

An AI co-pilot for Italian forfettario freelancers that ingests SDI invoices and expenses and answers 'how much tax and INPS do I owe and when' — with figures exportable for your commercialista to confirm.

Agents & AutomationDevelopers

wow 7pas 8.56w

#21

8.0

nis2-evidence

An AI NIS2 baseline assessor that reads your cloud/policy configs, auto-maps them to controls, and keeps a continuously-evidenced, board-ready posture for newly-in-scope EU mid-market companies.

AI Security & GovernanceDevelopers

wow 7pas 8.55w

#22

8.0

Reshape

A browser extension that rewrites the 10 most-hated SaaS UIs (Salesforce, HubSpot, Workday) in plain English — hide noise, surface your fields, one-click your workflow — and resyncs itself when the vendor ships changes.

AI Dev ToolsDevelopers

wow 9pas 7.05w

#23

8.0

GuardrailKit

Composable runtime guardrails that sit outside the model — input/output validation, policy, and PII redaction in five lines — with paid managed compliance packs for HIPAA, GDPR, and finance.

AI Security & GovernanceDevelopers

wow 7pas 8.54w

#24

8.0

LeadScout

An inbound-only intent radar that watches your chosen developer communities for people describing the exact pain your product solves, learning what a great match looks like from your feedback.

AI Dev ToolsDevelopers

wow 8pas 7.55w

#25

8.0

MCPGen

Point it at any OpenAPI spec or Postgres DB and it generates a typed, auth-scoped MCP server — then hosts it with a managed auth, rate-limit, and audit gateway in one command.

AI InfrastructureDevelopers

wow 8pas 7.56w

#26

8.0

ragbox

An EU-sovereign RAG stack in one Docker command — pgvector + local embeddings + a typed API — that also generates the data-residency and audit artifacts your client's compliance team demands.

AI InfrastructureDevelopers

wow 8pas 7.56w

#27

8.0

EvalRubric

Self-writing evals that point at your agent, generate a context-specific rubric from production traces, and gate every deploy in CI on regressions.

AI InfrastructureDevelopers

wow 8pas 7.54w

#28

8.0

AgentSLA

A status-page + uptime monitor for AI agents and LLM endpoints — tracks success rate, latency, drift and refusals — fronted by a public reliability leaderboard of models and agents.

AI InfrastructureDevelopers

wow 7pas 8.04w

#29

8.0

llmstxt-forge

The GEO score for any docs site: crawl it, generate a fresh llms.txt, then test model-by-model whether Claude, ChatGPT and Cursor actually cite you correctly — and monitor it over time.

AI InfrastructureDevelopers

wow 8pas 7.53w

#30

8.0

repo2agent

Point it at any GitHub repo and get an instant, deployable MCP server that lets Claude Code, Cursor, or your own chat widget answer questions grounded in that codebase — with citations.

AI Dev ToolsDevelopers

wow 8pas 7.56w

#31

8.0

ai-act-classifier

A free tool that tells any EU founder their AI system's risk tier and exact obligations, then sells a living dossier that re-checks those obligations and alerts when they change.

AI Security & GovernanceDevelopers

wow 8pas 7.02w

#32

7.9

OutcomeBill

Stripe-connected billing for outcome-priced AI products: define an outcome (ticket closed, report shipped), verify it from your agent logs, and invoice it — drop-in SDK, no billing team required.

Agents & AutomationDevelopers

wow 8pas 8.56w

#33

7.9

docling-cloud

A GDPR-safe, EU-hosted API and OSS pipeline that turns the worst PDFs, slides, and scans — tables, multi-column, handwritten — into clean, chunked, embedding-ready markdown for RAG.

AI Dev ToolsDevelopers

wow 7pas 8.55w

#34

7.9

SaaSKiller Kit

An open teardown kit + flagship app that replaces ONE painful legacy-SaaS workflow — vendor invoice reconciliation — with a 90%-automated, per-item-priced AI app, self-serve for SMBs.

Agents & AutomationDevelopers

wow 8pas 8.07w

#35

7.9

traceback

Open-source, OTel-native local trace viewer for AI agents that lets you time-travel and diff a failing run against the last good one — no cloud, no signup.

Agents & AutomationDevelopers

wow 9pas 7.05w

#36

7.9

sovereign-rag

A turnkey EU-region-only RAG starter kit you deploy in one command, plus a managed service that continuously attests and proves every byte stayed in the EU.

AI Dev ToolsDevelopers

wow 8pas 8.06w

#37

7.9

MCP Mission Control

Real-time mission control for agent fleets — live task board, cost-per-agent, and the moment an agent loops or stalls it pages you and can auto-kill the runaway before it burns tokens.

Agents & AutomationDevelopers

wow 7pas 8.05w

#38

7.9

agent-ledger

A self-hostable, tamper-evident audit log for AI agents that maps directly to EU AI Act Article 12 and produces the export your auditor accepts.

AI Security & GovernanceDevelopers

wow 7pas 8.05w

#39

7.9

SwitchCost

A migration agent that moves you off ONE specific legacy SaaS (start: Zendesk -> an AI-native helpdesk) and keeps both systems in sync during the transition — then expands one pair at a time.

Agents & AutomationDevelopers

wow 8pas 7.57w

#40

7.9

StatusAI

A live quality status page for AI chatbots: runs real synthetic conversations against your bot and publishes hallucination, refusal, latency and cost — not just up/down.

AI InfrastructureDevelopers

wow 9pas 7.05w

#41

7.9

MultiAgentBoard

A live control room for multi-agent systems that visualizes hand-offs and shared memory, and alerts you the moment an agent gets stuck, loops, or burns runaway cost.

AI InfrastructureDevelopers

wow 9pas 7.05w

#42

7.9

AgentLedger

A drop-in spend governor for AI agents: live per-agent cost, hard budget caps that actually stop runaway loops, and auto-routing each call to the cheapest capable model — priced on the savings it proves.

AI InfrastructureDevelopers

wow 7pas 8.04w

#43

7.9

VibeCheck

A PR-blocking GitHub Action that scores and gates AI-generated code, catching hallucinated imports, missing error handling, and security smells before they reach main.

AI Security & GovernanceDevelopers

wow 8pas 7.54w

#44

7.8

subagents.dev

The verified registry for coding-agent subagents and skills — install a benchmarked, trust-scored 'PR reviewer' or 'migration writer' into your repo with one command.

AI Dev ToolsDevelopers

wow 9pas 7.06w

#45

7.8

shipnotes

AI changelogs that are both human-grade and agent-readable — drafted from your merged PRs in your voice, published to a branded page, and emitted as a structured llms.txt feed agents can consume.

Agents & AutomationDevelopers

wow 7pas 8.03w

#46

7.8

BrowserPilot

Record one run of a flaky no-API web chore and get a reliable, self-healing computer-use script that re-anchors itself when the portal's UI shifts.

Agents & AutomationDevelopers

wow 7pas 7.55w

#47

7.8

golden-replay

Capture a real production agent failure from your OTel traces, scrub PII, and convert it into a deterministic golden test that runs in CI — so the bug can never silently return.

AI Security & GovernanceDevelopers

wow 8pas 7.05w

#48

7.8

SandboxRunner

A drop-in SDK + managed runner that executes your agent's generated code in a hardened ephemeral sandbox, with an escape-test suite and per-run audit log proving it never touched your host.

AI Security & GovernanceDevelopers

wow 8pas 7.05w

#49

7.8

agentsdotmd

One command turns any repo or docs site into an agent-ready llms.txt + hosted MCP server, then continuously monitors and alerts you when AI agents misread your docs — backed by a public Agent-Readability Index.

AI Dev ToolsDevelopers

wow 8pas 7.04w

#50

7.8

DocsToAgent

Make your docs answerable by every AI: one command ships an llms.txt, an AEO audit, and an embeddable support agent that cites the source page.

AI Security & GovernanceDevelopers

wow 8pas 7.05w

#51

7.8

ai-act-changelog

A subscribe-able AI changelog of every EU AI Act / GPAI / standards update that tells you 'what changed for YOUR risk tier and obligations' — not 200 pages.

AI Security & GovernanceDevelopers

wow 8pas 7.02w

#52

7.7

VoiceKit OSS

An open-source real-time voice-agent starter (turn detection, barge-in, tool calls) on your own keys — plus a hosted 'voice eval & latency dashboard' that proves your agent actually feels human.

OSS & Developer ExperienceDevelopers

wow 9pas 7.06w

#53

7.7

doc-residency-vault

A self-hostable audit layer that wraps your existing EU vector DB so every embedding chunk carries a signed residency + access proof for regulators.

AI InfrastructureDevelopers

wow 8pas 7.56w

#54

7.7

ChangelogAI

Turns merged PRs into a polished public changelog AND auto-drafted launch posts for X/LinkedIn — your ship-in-public loop on autopilot.

Creator & ContentDevelopers

wow 7pas 8.04w

#55

7.7

PRReviewer

A self-hosted AI PR reviewer that turns your CLAUDE.md/AGENTS.md into hard, deterministic gates—and only spends LLM tokens on the judgment calls.

AI Dev ToolsDevelopers

wow 8pas 7.05w

#56

7.7

EvalForge

Open-source 'pytest for agents' — write task-based evals as code, run them in CI, and block the PR when your agent's reliability regresses, with a shareable reliability badge.

AI Dev ToolsDevelopers

wow 8pas 7.06w

#57

7.7

SkillFile

Turn a support team's scattered SOPs (Zendesk macros, Notion runbooks, Slack threads) into a versioned, portable 'skills file' that drops into Fin/Decagon/your own agent so it acts like a trained rep on day one.

Vertical AI SaaSDevelopers

wow 8pas 7.06w

#58

7.7

ContextPack

Measures and optimizes what your coding agent actually sees: one command produces a token-budgeted context pack AND an open benchmark score for how much that pack improves real agent task success.

Agents & AutomationDevelopers

wow 8pas 7.04w

#59

7.7

OneClickAgent

One command wraps any agent in a production guardrail: an eval gate that blocks quality regressions and a hard cost cap that prevents runaway spend, with tracing built in.

AI InfrastructureDevelopers

wow 8pas 7.07w

#60

7.7

RubricBench

Run a trustworthy, task-specific agent benchmark in one click — public commons plus private suites that re-run on every model release.

AI Dev ToolsDevelopers

wow 8pas 7.04w

#61

7.7

trending-radar

A daily, AI-curated GitHub opportunity radar for builders — fast-rising repos filtered to your stack, each with a sharp 'why it matters' and a 'what to build next' gap take, delivered to email/RSS/Slack with an open data feed.

AI Dev ToolsDevelopers

wow 8pas 7.03w

#62

7.7

OpenJudge

A calibrated LLM-as-judge with a free reliability audit that shows how biased and miscalibrated your current judge is — then fixes it with confidence scores and drift monitoring.

AI InfrastructureDevelopers

wow 7pas 7.04w

#63

6.6

eu-eval-local

A self-hosted, GDPR-safe LLM eval + observability harness for EU teams who can't send prompts to US SaaS — shipping a signed data-residency attestation legal will actually accept.

AI Dev ToolsDevelopers

wow 7pas 6.06w

#64

6.2

AgentBudget

A spend control plane for agent fleets: hard per-task budgets, a runaway kill-switch, and team-wide cost allocation so no agent can ever burn an unexpected bill.

Agents & AutomationDevelopers

wow 6pas 6.04w

#65

6.1

pii-fence

A drop-in middleware that detects and redacts PII before any LLM call and emits a tamper-evident, EU-residency audit dossier your DPO can hand an auditor.

AI Security & GovernanceDevelopers

wow 6pas 6.04w

#66

6.0

PolicyAgent

Connects to your stack, auto-detects new AI sub-processors and data flows, and keeps your privacy policy and AI-Act disclosures truthful and current — hosted at a URL you never have to touch.

AI Security & GovernanceDevelopers

wow 6pas 6.05w

#67

6.0

eu-model-router

An EU-sovereign LLM gateway that returns a cryptographically signed residency receipt for every request, proving no call ever left the EU.

AI InfrastructureDevelopers

wow 6pas 6.05w

#68

6.0

aiact-evidence

An open-source TypeScript SDK that auto-generates EU AI Act Article 11 technical documentation and tamper-evident Article 12 event logs from your running AI app, with a hosted 10-year vault and auditor-shareable dashboard.

AI Security & GovernanceDevelopers

wow 6pas 6.04w

#69

5.5

DriftWatch

Monitors LLM providers for silent model/behavior changes—runs canary prompts, scores semantic drift, and alerts you the moment 'the same prompt' starts answering differently.

AI InfrastructureDevelopers

wow 5pas 5.05w

#70

5.4

AgentInbox

A human-in-the-loop approval inbox and audit trail that gives your autonomous agents provable human oversight — one-tap sign-off in Slack and an AI-Act-ready compliance log.

AI Security & GovernanceDevelopers

wow 5pas 5.05w

#71

5.4

multiagent-canvas

An open-source visual observer for multi-agent runs — ingest a CrewAI/AutoGen/LangGraph trace and get a live, replayable canvas showing who talked, what tools fired, and exactly where it stalled.

Agents & AutomationDevelopers

wow 6pas 4.07w

#72

5.3

prompt-injection-eu

An open-source agentic red-team + guardrail suite with a public model-robustness leaderboard that maps every finding to EU AI Act Art.-15 robustness/cybersecurity evidence.

AI Security & GovernanceDevelopers

wow 6pas 5.05w

#73

5.3

ActAudit

Record any computer-use agent run as a signed, replayable audit trail — share a free replay link or store it in a tamper-evident compliance vault.

AI Security & GovernanceDevelopers

wow 6pas 4.55w

#74

5.3

RAGStudio

Open-source RAG bench with a public 'best config 2026' leaderboard — upload your own docs to get a measured retrieval-quality score, chunking/embedder/reranker comparison, and an exportable tuned pipeline.

AI InfrastructureDevelopers

wow 6pas 4.06w

#75

5.2

AgentTrace

Self-hostable single-binary tracer for autonomous agents: replay any run as a flamegraph that auto-flags loops, cost spikes, and bad tool calls — with shareable trace permalinks.

Agents & AutomationDevelopers

wow 6pas 4.06w

#76

5.2

eaa-scanner

Open-source CI-native accessibility scanner that maps WCAG 2.2 failures to specific European Accessibility Act obligations, opens AI-generated fix PRs, and produces the audit dossier.

AI Security & GovernanceDevelopers

wow 5pas 5.04w

#77

5.1

agent-cost-guard

A self-hosted enforcing proxy for AI agents that sets hard per-tenant budget caps, kills runaway spend in real time, and emits chargeback-ready cost reports for finance.

Agents & AutomationDevelopers

wow 5pas 5.04w

#78

5.0

annexiv-from-repo

A GitHub App that reads your repo and drafts auditable EU AI Act Annex-IV technical documentation grounded in cited code/config lines, then keeps it synced as your code changes.

AI Security & GovernanceDevelopers

wow 4pas 5.05w

#79

4.8

FormToAgent

An AI demo-request agent for B2B SaaS that interviews each visitor, scores fit using your own closed-won history, and books qualified meetings straight into your CRM.

Vertical AI SaaSDevelopers

wow 4pas 5.05w

#80

4.7

mcp-doctor

The security-conformance scanner and trust registry for MCP servers — prove your server is injection-safe and schema-clean before vendors let agents connect.

AI Security & GovernanceDevelopers

wow 5pas 4.05w

#81

4.3

AgentDocs

Auto-generate, host, and continuously tune agent-facing MCP tool docs from your OpenAPI/code — using real agent-call telemetry to drive call-correctness up and keep descriptions in sync as your API drifts.

AI Dev ToolsDevelopers

wow 4pas 4.04w

#82

4.3

eurlex-mcp

An MCP server that makes the EU AI Act + DORA directly queryable with paragraph-level citations from any AI client, plus paid change-alerts when the regulation that affects you is amended.

AI Security & GovernanceDevelopers

wow 4pas 4.04w

#83

4.3

EvalKit

Open-source, one-command LLM eval harness that turns your prod traces into a green/red regression gate for prompts and agents — runs in CI, no platform to adopt.

AI InfrastructureDevelopers

wow 4pas 4.05w

#84

4.1

agentwall

A local-first firewall for AI coding agents — mask secrets, block destructive shell calls, and catch prompt injection on the wire, with a managed detector feed and team audit log.

AI Security & GovernanceDevelopers

wow 4pas 3.56w

#85

4.1

TokenLens

A cost x-ray for agent runs that not only shows which tools, prompts, and retries burn your budget but auto-recommends fixes and alerts before the bill spikes.

AI InfrastructureDevelopers

wow 4pas 3.03w

#86

4.0

RAGAudit

Point it at your RAG pipeline and get a brutal scorecard — retrieval recall, hallucination rate, stale-doc detection — with the exact chunking/retrieval fix, then it guards every deploy against regressions on your golden set.

Data & RAGDevelopers

wow 4pas 4.05w

#87

4.0

mcp-sandbox

Scan and sandbox any community MCP server before you trust it: a one-command safety report plus an isolated run-mode with network/file allowlists, backed by a public MCP safety directory.

AI Security & GovernanceDevelopers

wow 4pas 3.55w

#88

3.9

injectbench

An open red-team benchmark, public leaderboard, and CI gate that fires 500+ evolving prompt-injection payloads at your agent or MCP server and scores how many it survives — with a README badge.

AI Security & GovernanceDevelopers

wow 4pas 3.05w

#89

3.8

FallbackRouter

Drop-in OSS router that auto-fails over refused or geo-blocked model calls, powered by a live cross-provider model-health dataset you subscribe to.

AI InfrastructureDevelopers

wow 4pas 3.03w

#90

3.8

PromptArmor

An open-source prompt-injection firewall plus public attack benchmark that scores and blocks tool-call exploits before they fire.

AI Security & GovernanceDevelopers

wow 4pas 3.04w

#91

3.6

SkillSmith

Author, test, and version Claude Skills like real software — scaffolding, an eval harness, and CI gates that block a skill from shipping if it regressed.

AI Dev ToolsDevelopers

wow 4pas 3.03w

#92

3.5

vat-moss-agent

An MCP/SDK that computes correct EU VAT (per-country rates, reverse charge, OSS) for digital sales and exports filing-ready OSS reports — backed by a free, always-current public rate table.

AI Dev ToolsDevelopers

wow 3pas 3.55w

#93

3.4

RedTeamGPT

A CI-installable jailbreak suite that attacks your AI agent on every deploy and blocks the release if it leaks its prompt, breaks policy, or can be hijacked.

AI Security & GovernanceDevelopers

wow 3pas 3.06w