Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
125 changes: 125 additions & 0 deletions launch/enterprise-outbound-cadence-and-contracts.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,125 @@
# Enterprise Outbound Motion & Private VPC Pilot Contracts ($25k–$50k/yr)

This document contains the complete outbound sales cadences, legal contract agreements, and compliance trust package for closing $25k–$50k/yr Enterprise Private VPC contracts.

---

## 🎯 Part 1: Dual-Persona Outbound Sales Cadence

### Persona A: VP of Engineering / Head of FinOps
**Pain Point**: Exploding LLM token bills from multi-agent coding loops and long CI build times in monorepos.

#### Email 1 (Day 1): The Empirical FinOps Angle
**Subject**: Cutting 98% of your autonomous agent prompt token bill
**Body**:
> Hi {{firstName}},
>
> Noticed your team at {{company}} is standardizing on autonomous coding agents (Claude Code / Cursor / Windsurf).
>
> Most engineering orgs running 50+ agents are seeing LLM inference bills surge past $15,000–$40,000/mo because agents repeatedly dump 40k+ tokens of raw tool schemas and git context into every turn.
>
> We built **SkillPrism** to stack JIT semantic tool pruning with native Anthropic/OpenAI prompt caching. Across our production fleets, we consistently achieve a **98%+ billable prompt token reduction** ($40k context compressed to 25 effective billable tokens) with sub-10ms latency.
>
> I modeled your team's potential savings on our live FinOps ROI Calculator here:
> https://ritualdev.in/finops-calculator
>
> Would you be open to a 15-minute briefing on deploying our dedicated VPC cluster in your cloud?
>
> Best,
> Alex Rivera
> Enterprise Solutions, RitualDev Lab
> [email protected]

#### Email 2 (Day 4): The Benchmark Proof & Monorepo Cache
**Subject**: Re: Cutting 98% of your autonomous agent prompt token bill
**Body**:
> Hi {{firstName}},
>
> Following up with the empirical benchmark: In our latest evaluation across 500 standardized workloads, SkillPrism saved $34,200/mo on a 5,000 req/day agent fleet while exporting full OpenTelemetry FinOps metrics (`compute_dollars_saved`, `tokens_compacted_total`) directly into Datadog.
>
> In addition, our unified suite bundles **Poly Cloud Cache**, accelerating monorepo builds by 6.4x via shared remote S3/R2 caching.
>
> We offer a turnkey **60-Day Dedicated VPC Pilot ($25,000/yr)** where 100% of pilot fees are credited towards your annual license.
>
> Open to seeing a live architecture demo on Thursday?
>
> Best,
> Alex

---

### Persona B: Chief Information Security Officer (CISO) / VP DevSecOps
**Pain Point**: Autonomous agents executing unchecked terminal commands, leaking AWS/GCP IMDS metadata, and running unconstrained SQL queries against production databases.

#### Email 1 (Day 1): The Kernel Sandbox & Zero-Rollback Angle
**Subject**: Zero-overhead eBPF sandboxing for {{company}}'s AI coding agents
**Body**:
> Hi {{firstName}},
>
> As {{company}} evaluates autonomous coding agents with shell and database privileges, how are you enforcing runtime guardrails at the OS syscall and SQL wire level?
>
> Secondary LLM guardrails (LlamaGuard, NeMo) add 300ms+ latency and cannot intercept raw OS syscalls (`clone3`, `unshare`) or prevent destructive database migrations once TCP packets leave the agent.
>
> We built **AgentShield & DBShield**:
> - **AgentShield**: Linux eBPF CO-RE runtime isolation with **<0.10ms P99 overhead**, strictly blocking path traversal, environment sniffing, and cloud IMDS (`169.254.169.254`) egress.
> - **DBShield**: Wire-level SQL proxy executing **guaranteed immediate ROLLBACK** on unconstrained `DROP TABLE`, `TRUNCATE`, or bulk `UPDATE` statements.
>
> In our 500-vector Agent Security Benchmark (ASB), AgentShield achieved a **100% block rate**, compared to 41.8% for LlamaGuard 3:
> https://ritualdev.in/benchmarks
>
> We provide pre-compiled SOC 2 Type II, ISO 27001, and SIG Lite 2026 trust packages for deployment in your air-gapped VPC.
>
> Do you have 15 minutes this week to review the architecture?
>
> Best,
> Marcus Vance
> Head of Security Solutions, RitualDev Lab
> [email protected]

---

## 📜 Part 2: Enterprise Commercial Pilot Agreement (SOW)

### Commercial Pilot Statement of Work (SOW #RIT-2026-VPC)
**Effective Date**: [EffectiveDate]
**Provider**: RitualDev Lab Inc. ("Provider")
**Customer**: [CustomerName] ("Customer")

#### 1. Scope & Architecture
Provider shall deploy a single-tenant, air-gapped **RitualDev Enterprise Governance Suite** inside Customer's designated cloud account (AWS, GCP, Azure, or on-premises Kubernetes). The deployment includes:
- **AgentShield Kernel DaemonSet**: Linux eBPF CO-RE syscall containment and persistent Zygote process pool.
- **DBShield Wire Defense Proxy**: Detached dry-run SQL engine with automatic rollback.
- **SkillPrism FinOps Gateway**: 98%+ prompt token compaction with OpenTelemetry exporter.
- **Poly Cloud API**: High-throughput remote build cache service.

#### 2. Service Level Agreements (SLAs)
- **Runtime Latency**: Sandbox intercept overhead shall not exceed **0.10ms P99**.
- **Data Sovereignty**: 100% air-gapped. Zero prompts, source code, or database rows shall leave Customer's VPC boundary.
- **Uptime**: 99.9% uptime during the 60-day pilot window.

#### 3. Commercial Fees & Conversion Terms
- **Pilot Term**: Sixty (60) days from deployment completion.
- **Pilot Investment**:
- **Option 1 (Team VPC Pilot)**: $25,000 USD (up to 100 concurrent agents).
- **Option 2 (Enterprise Air-Gapped VPC Pilot)**: $50,000 USD (unlimited agents, CMEK encryption, custom SIEM integration).
- **Annual Conversion Credit**: One hundred percent (100%) of the Pilot Investment shall be credited directly toward Customer's Year 1 Master Services Agreement upon contract execution before Day 60.

#### 4. Compliance & Trust Package
Provider certifies that the software complies with:
- **SOC 2 Type II**: CC6.1, CC6.6, CC6.8, CC7.1, CC7.2.
- **ISO/IEC 27001:2022**: Annex A.8.20, A.8.24, A.8.28.
- **SIG Lite 2026**: Fully pre-filled questionnaire provided upon signature.

**IN WITNESS WHEREOF**, the parties have executed this Pilot Agreement.

**Customer**:
Signature: __________________________
Name: _______________________________
Title: ______________________________
Date: _______________________________

**RitualDev Lab Inc.**:
Signature: __________________________
Name: _______________________________
Title: Chief Commercial Officer
Date: _______________________________
37 changes: 37 additions & 0 deletions launch/week10-skillprism-finops-launch.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,37 @@
# Week 10 Launch Pack: SkillPrism Token FinOps

## 🚀 Show HN Submission

**Title**: Show HN: SkillPrism – 98% Prompt Token Compaction with Stacked Prompt Caching & OTel Metrics

**URL**: https://ritualdev.in/finops-calculator (or https://github.com/RitualDev-Lab/SkillPrism)

**Body**:
Hey HN!

As autonomous agent workflows become multi-step and multi-agent, token bills explode exponentially. Agents routinely dump 30,000 to 80,000 tokens of tool schemas, AST snippets, and environment context into every single turn.

We built **SkillPrism** to solve LLM token FinOps from first principles:

1. **JIT Semantic Pruning**: Identifies the exact tool schemas and context slices required for the active turn, pruning 80%+ of unused token bloat.
2. **Native Prompt Cache Stacking**: Injects Anthropic ephemeral `cache_control` breakpoints and OpenAI deterministic prefix sorting on top of the pruned context.
3. **98%+ Real-World Reduction**: Compresses typical 40,000-token tool contexts down to **25 effective billable tokens** on cache hits.
4. **OpenTelemetry FinOps Exporter**: Exports Prometheus metrics and OTLP JSON (`tokens_compacted_total`, `compute_dollars_saved`, `p99_latency_ms`) directly to Datadog and Grafana.
5. **HA Redis Sentinel**: 3-AZ clustering for sub-10ms semantic routing across agent clusters.

We also built an interactive FinOps ROI calculator where you can model your exact agent fleet savings:
https://ritualdev.in/finops-calculator

CLI 1-liner:
```bash
npx @ritualdev-lab/skillprism init
```

Check it out on GitHub: https://github.com/RitualDev-Lab/SkillPrism

---

## 🏆 Product Hunt Launch Kit

**Tagline**: 98%+ prompt token reduction with stacked caching & FinOps telemetry
**Target Persona**: VPs of Engineering, FinOps leaders, and AI agent developers.
29 changes: 29 additions & 0 deletions launch/week11-devshelf-universal-credits-launch.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,29 @@
# Week 11 Launch Pack: DevShelf & Universal MCP Credits

## 🚀 Show HN Submission

**Title**: Show HN: DevShelf & Universal MCP Credits – 575+ Tools with 1-Click Monetization & Security Badges

**URL**: https://ritualdev.in/mcp-security (or https://github.com/RitualDev-Lab/DevShelf)

**Body**:
Hey HN!

The Model Context Protocol ecosystem has grown to hundreds of community tools, but two major problems remain:
1. **Security**: Our audit of the Top 100 MCP servers revealed that 26% expose host terminals or cloud IMDS endpoints without basic sandboxing.
2. **Monetization & Billing**: Tool creators have no standardized way to charge agents micropayments for API or compute usage.

Today we're releasing the dual solution:

1. **DevShelf Verified Security & Uptime Badges ($49/mo)**: Continuous AST taint-flow scanning via WebSecAuditor that certifies MCP servers with cryptographically signed verification badges.
2. **Universal Pre-Paid MCP Credits via `mcp-monetize`**: An edge-accelerated billing clearinghouse running on Cloudflare Durable Objects. Developers monetize any MCP server with 3 lines of code while users pay with a single universal balance across 575+ developer tools.

Explore the Top 100 MCP Vulnerability Report: https://ritualdev.in/mcp-security

Explore the catalog: https://ritualdev.in/devshelf

---

## 🏆 Product Hunt Launch Kit

**Tagline**: The secure, monetized developer shelf for autonomous AI agent tools.
58 changes: 58 additions & 0 deletions launch/week9-agentshield-dbshield-launch.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,58 @@
# Week 9 Launch Pack: AgentShield & DBShield

## 🚀 Show HN Submission

**Title**: Show HN: AgentShield & DBShield – Sub-0.10ms Linux eBPF Sandbox & Instant SQL Rollback for AI Agents

**URL**: https://ritualdev.in/benchmarks (or https://github.com/RitualDev-Lab/AgentShield)

**Body**:
Hey HN! We're the team behind RitualDev Lab.

As autonomous AI coding agents (Claude Code, Windsurf, Cursor, OpenHands) gain access to local shells, git workspaces, and production databases, standard prompt-level guardrails are falling short:
1. They add 200ms–600ms latency to every single tool step.
2. They cannot intercept raw OS syscalls (e.g. `clone3`, `unshare`, `/proc/kcore` memory dumps).
3. They cannot prevent destructive SQL mutations once a database TCP packet leaves the agent.

To solve this, we built and open-sourced **AgentShield** and **DBShield**:

- **AgentShield**: A sub-0.10ms Linux eBPF CO-RE runtime sandbox. By pooling pre-forked Zygote processes and enforcing seccomp/Landlock filesystem bounds, it intercepts unauthorized syscalls and network egress (including AWS/GCP IMDS `169.254.169.254`) with **<0.09ms P99 overhead**.
- **DBShield**: A wire-level SQL defense proxy. It intercepts database streams in detached lock-free dry-runs. When an agent emits an unconstrained `DROP TABLE`, `TRUNCATE`, or bulk `UPDATE` without `WHERE`, DBShield executes an immediate zero-lock **ROLLBACK** before data is committed.

We also benchmarked both against 500 standardized attack vectors in our open-source Agent Security Benchmarks (ASB) suite:
- AgentShield + DBShield: **100% block rate** at **<0.09ms P99 latency**.
- LlamaGuard 3 (8B): **41.8% block rate** at **314ms latency**.
- NeMo Guardrails: **53.6% block rate** at **217ms latency**.

You can run the 500-vector benchmark locally:
```bash
git clone https://github.com/RitualDev-Lab/awesome-ai-agent-security.git
cd awesome-ai-agent-security/benchmarks
npm test
```

Try the CLI in 10 seconds:
```bash
npx @ritualdev-lab/agentshield init
```

We would love your thoughts on our eBPF and wire proxy architecture!

---

## 🏆 Product Hunt Launch Kit

**Tagline**: Zero-overhead Linux kernel sandbox & wire SQL rollback for AI agents
**Pricing**: Free Open Source Community ($0) / Enterprise Air-Gapped Private VPC ($25k–$50k/yr)

**First Maker Comment**:
Hey hunters! 👋

We built AgentShield and DBShield because we love autonomous coding agents, but we were terrified of running them with full shell and database privileges on our production machines.

Instead of adding slow LLM classifiers that can be jailbroken with clever roleplay, we went down to the Linux kernel and TCP wire:
1. **eBPF CO-RE Probes**: Block unauthorized filesystem traversal and IMDS token theft at the kernel boundary.
2. **Wire Rollback**: Guarantee zero data loss when agents hallucinate database migrations.
3. **<0.10ms Latency**: 1000x faster than secondary LLM guardrails.

All core components are open source on GitHub. Check out the live benchmark matrix at https://ritualdev.in/benchmarks!
Loading
Loading