What is ThumbGate CLAUDE.md?

Project instructions for ThumbGate, a service that acts as an infrastructure firewall around AI coding agents. They define the production service, repository, package, deployment process, and operating responsibilities.

Use them when updating ThumbGate, synchronizing its version, managing GitHub changes, or checking how changes reach the Railway-hosted production service.

They give agents the fixed project facts and deployment rules needed to make coordinated changes without asking the owner to perform routine steps manually.

How IgorGanapolsky/ThumbGate configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to IgorGanapolsky/ThumbGate. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/IgorGanapolsky/ThumbGate/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/IgorGanapolsky/ThumbGate

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/igorganapolsky/thumbgate/claude-md/github.svg)](https://agentmods.dev/instructions/igorganapolsky/thumbgate/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/igorganapolsky/thumbgate/claude-md"><img src="https://agentmods.dev/badge/instructions/igorganapolsky/thumbgate/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/igorganapolsky/thumbgate/claude-md"><img src="https://agentmods.dev/badge/instructions/igorganapolsky/thumbgate/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 7,040 This file is loaded in full into every session.

When invoked 7,040 The same file — it is already loaded in full.

Security scan A 1 finding. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.07040	$0.07040
Opus 5	$0.03520	$0.03520
Sonnet 5	$0.01408	$0.01408
Haiku 4.5	$0.00704	$0.00704

Measured 4d ago against content hash 192f6e7b2238, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

ThumbGate CLAUDE.md scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

curl -s https://thumbgate-production.up.railway.app/health | grep "\"version\":\"${EXPECTED_VERSION}\""

CLAUDE.md · 445 lines

How it starts

The opening of the file, as written. The whole thing — 445 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — ThumbGate (`thumbgate`)

Constants

PROD_URL    = https://thumbgate-production.up.railway.app
REPO        = IgorGanapolsky/ThumbGate
CORE_REPO   = IgorGanapolsky/ThumbGate-Core
NPM_PKG     = thumbgate
VERSION     = package.json  (source of truth: scripts/sync-version.js propagates release surfaces)
DEPLOY      = Railway auto-deploys from main via Docker (2-5 min rebuild)

Autonomy Directive

You are the CTO. Igor Ganapolsky is your CEO. Execute autonomously: branch, commit, push, PR, merge, deploy. Never tell the CEO to run a command — run it yourself. Never leave a PR open when CI passes and threads are resolved.

Never tell the CEO to do anything manually. If something needs doing, do it yourself.

For PR-management sessions, GitHub Issues and Actions are mandatory evidence surfaces. Reconcile Linear AI, ~/Documents/AI-Agent-Sync, and one non-interactive standalone copilot -p advisory when those local integrations are available. Otherwise record their absence and continue with deterministic gh, git, and repository checks, which remain authoritative.

What This Repo Is

ThumbGate: infrastructure firewalls for AI coding agents. Captures feedback → promotes to memory → generates prevention rules → blocks known-bad tool calls via PreToolUse hooks.

Not traditional model-training feedback optimization. It is context engineering + enforcement.

Stack: Node.js >=18.18.0, SQLite+FTS5 lesson DB, Thompson Sampling, LanceDB vectors, ContextFS context assembly.

Canonical Product Scope

ThumbGate is the only active source of truth. Use IgorGanapolsky/ThumbGate, npm package thumbgate, and https://thumbgate-production.up.railway.app for repo, package, launch, GPT Actions, and production API references.

Do not use mcp-memory-gateway, rlhf, old RLHF-loop paths, or similarly named local worktrees/remotes as active product context. Those strings are legacy migration aliases only when cleanup code or tests intentionally remove old config keys.

Read the full file on GitHub · 445 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

4d ago Changed · +14 lines · +156 tokens per session 192f6e7b2238
9d ago First seen · 431 lines · 6,884 tokens per session scan A 535f72587b1f

Subscribe to this mod's changes

ThumbGate CLAUDE.md is an instructions file published in the GitHub repository IgorGanapolsky/ThumbGate (26 stars, last pushed today), licensed MIT. It adds 7,040 tokens to every session, about $0.0352 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

AI Safety Guardrails agent-reliability ai-agents developer-tools thompson-sampling

ThumbGate: Instructions file for Claude Code

Borrowing it

Wrote this? Show the measurements

What it costs to keep this loaded

Grade A, and why

Makes network callslowCapability

How it starts

CLAUDE.md — ThumbGate (`thumbgate`)

Constants

Autonomy Directive

What This Repo Is

Canonical Product Scope

What this file has done since we first saw it

Other instructions, from other repositories

plan-forge blazor-fluent-ui.instructions.md

plan-forge landing-zone.instructions.md

plan-forge bicep.instructions.md

plan-forge policy.instructions.md

plan-forge api-patterns.instructions.md

plan-forge messaging.instructions.md

ThumbGate: Instructions file for Claude Code

Borrowing it

Wrote this? Show the measurements

What it costs to keep this loaded

Grade A, and why

Makes network callslowCapability

How it starts

CLAUDE.md — ThumbGate (thumbgate)

Constants

Autonomy Directive

What This Repo Is

Canonical Product Scope

What this file has done since we first saw it

Other instructions, from other repositories

plan-forge blazor-fluent-ui.instructions.md

plan-forge landing-zone.instructions.md

plan-forge bicep.instructions.md

plan-forge policy.instructions.md

plan-forge api-patterns.instructions.md

plan-forge messaging.instructions.md

CLAUDE.md — ThumbGate (`thumbgate`)