Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/matthiastjong/shellgate-legacy/agents-mdgit clone --depth 1 https://github.com/matthiastjong/shellgate-legacyWhat 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 | $0.02790 | $0.02790 |
| Opus 5 | $0.01395 | $0.01395 |
| Sonnet 5 | $0.00558 | $0.00558 |
| Haiku 4.5 | $0.00279 | $0.00279 |
Grade B, and why
shellgate-legacy AGENTS.md scanned grade B 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 3d 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
Shellgate exposes all agent-facing functionality as an MCP server at `POST /mcp` using Streamable HTTP transport. Claude Code connects via `mcpServers` config in `~/.claude/settings.json`. How it starts
The opening of the file, as written. The whole thing — 233 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Development Guidelines
Architecture
Shellgate is a SvelteKit monolith — dashboard + API + gateway in one process.
Layers
src/lib/server/services/ ← All business logic lives here
src/lib/server/utils/ ← Pure helpers (CIDR, URL validation, etc.)
src/routes/api/ ← Admin API (thin wrappers calling services)
src/routes/(app)/ ← Dashboard pages (call services directly, no HTTP to self)
src/routes/gateway/ ← Agent-facing: HTTP proxy to upstream APIs
src/routes/ssh/ ← Agent-facing: SSH command execution
src/routes/discovery/ ← Agent-facing: list accessible targets
src/routes/bootstrap/ ← Agent-facing: full session-start context (REST)
Key services
| Service | Purpose |
|---|---|
targets |
CRUD for API, SSH, and email targets |
tokens |
API key generation, hashing, revocation |
permissions |
Token ↔ target access control |
auth-methods |
Credentials stored per target (bearer, basic, custom_header, ssh_key, imap_smtp) |
gateway |
Proxy logic: resolve target, inject auth, forward request |
ssh |
SSH command execution via ssh2 |
mail |
Email operations via ImapFlow (IMAP) + Nodemailer (SMTP) |
audit |
Request logging to audit_logs table |
users |
Dashboard user management (email + password) |
webhook-endpoints |
CRUD for incoming webhook endpoint registrations |
webhook-events |
Event creation, polling, ACK, cleanup |
connected-accounts |
OAuth account linking, managed target provisioning, token refresh |
Data model
tokens ──┐
├── token_permissions (unique: token + target)
targets ──┘
│
└── target_auth_methods (one default per target)
tokens ──── webhook_endpoints (one token can have multiple endpoints)
│
└── webhook_events (pending/delivered/expired)
connected_accounts ──── targets (via connected_account_id FK, cascade delete)
users (dashboard login)
audit_logs (every gateway + SSH + mail request)
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.
- 3d ago First seen · 233 lines · 2,790 tokens per session scan B 44b04c85a89e
shellgate-legacy AGENTS.md is an instructions file published in the GitHub repository matthiastjong/shellgate-legacy (15 stars, last pushed 2mo ago), licensed MIT. It adds 2,790 tokens to every session, about $0.0140 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
zentinel AGENTS.md
Instructions for zentinelproxy/zentinel, covering zentinel, philosophy (north star), architecture, crates and core crates.
mockserver-monorepo AGENTS.md
AGENTS.md instructions for mock-server/mockserver-monorepo, covering mockserver — agent instructions, instruction priority, project overview, local development environment and project documentation.
open-pencil copilot-instructions.md
Copilot instructions for open-pencil/open-pencil, covering openpencil copilot instructions, pull request requirements, validation and code conventions.
openscreen AGENTS.md
Instructions for getopenscreen/openscreen, covering agents.md, setup commands, development principles, project layout and code style.
openstatus AGENTS.md
AGENTS.md instructions for openstatusHQ/openstatus, covering agents.md, verify your change, toolchain, architecture and tests.
dario CLAUDE.md
Instructions for askalf/dario, covering dario — notes for claude code (and other llm coding agents), commits and prs, style, scope and when in doubt, ask.