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/agent-hellboy/mcp-runtime/agents-mdgit clone --depth 1 https://github.com/Agent-Hellboy/mcp-runtimeWhat 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 | $0.02649 | $0.02649 |
| Opus 5 | $0.01324 | $0.01324 |
| Sonnet 5 | $0.00530 | $0.00530 |
| Haiku 4.5 | $0.00265 | $0.00265 |
Grade A, and why
mcp-runtime AGENTS.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 2d 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.
Prereqs: Docker, Kind, `kubectl`, `curl`, `jq`, Python 3, Go. How it starts
The opening of the file, as written. The whole thing — 161 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — developer and AI-agent guide
This file is the onboarding index for the MCP Runtime repo. It complements README.md with where to look, build/CI conventions, and pointers to focused runbooks. Prefer repo source (README, CRDs, v1alpha1 types) over generic Kubernetes or MCP advice.
Operational detail lives in .codex/skills/ (symlinked from .claude/skills). Load the skill that matches your task instead of re-reading long checklists here.
| Task | Skill |
|---|---|
| Kind contributor cluster bring-up | qa-cluster-bringup |
| Local URLs, API keys, test logins | mcp-runtime-local-dev |
| Grants, sessions, MCP JSON-RPC | mcp-runtime-governance |
| Cluster failures (401, ingress, registry, pulls) | mcp-runtime-troubleshooting |
| Public domain, TLS, ACME, prod hostnames | mcp-runtime-platform-public |
| Public k3s deploy scripts | k3s-public-ops |
| Real-cluster QA sweeps | qa-e2e-operations, qa-e2e-security, qa-e2e-ui, … |
| Release / ship readiness | release-readiness |
| API / CRD / CLI design & "is this good design?" review | design-principles |
| Codebase navigation | graphify (when graphify-out/ exists) |
Repository map (where to look)
| Area | Path | Notes |
|---|---|---|
| User-facing CLI | cmd/mcp-runtime/, internal/cli/root/, internal/cli/<command>/, internal/cli/core/ |
Cobra routing; setup, status, registry, server, access, … |
| Agent adapters | internal/cli/adapter/, internal/agentadapter/, services/runtime-api/internal/runtimeapi/adapter*.go |
adapter proxy/stdio use issued sessions; adapter enroll submits a local-key CSR for enterprise mTLS |
| Operator | cmd/operator/, internal/operator/ |
MCPServer reconciliation, ingress (ingressClass default traefik), gateway |
| API & CRD types | api/v1alpha1/, config/crd/bases/ |
Source of truth for object shapes |
| Access and policy | pkg/access/, pkg/policy/ |
Grant/session helpers; gateway policy contract |
| Control-plane / K8s | pkg/controlplane/, pkg/k8sclient/, pkg/kubeworkload/, pkg/manifest/, pkg/metadata/ |
MCPServer ops, manifests, registry resolution |
| Sentinel packages | pkg/events/, pkg/clickhouse/, pkg/serviceutil/, pkg/sentinel/ |
Events, analytics, service utilities |
| Sentinel services | services/platform-api, services/runtime-api, services/analytics-api, services/ui, services/ingest, services/processor, services/mcp-gateway, … |
Separate go.mod where present; Go 1.26 for shared imports |
| Samples / install YAML | examples/workspace-assistant-mcp/, k8s/, config/ |
Demo server; overlays and CRDs |
| Team isolation | docs/multi-team.md |
Namespaces, RBAC, ingress watch scope |
| Deployment targets | docs/deployment-targets.md, docs/k3s-on-prem-cluster.md |
Before distribution-specific runbooks |
| E2E | test/e2e/, test/integration/ |
Kind script; envtest integration |
| Agent skills | .codex/skills/, .claude/skills → ../.codex/skills |
Canonical skills tree |
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.
- 2d ago First seen · 161 lines · 2,649 tokens per session scan A 2130df679c85
mcp-runtime AGENTS.md is an instructions file published in the GitHub repository Agent-Hellboy/mcp-runtime (5 stars, last pushed 8d ago), licensed Apache-2.0. It adds 2,649 tokens to every session, about $0.0132 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 2026-08-31.
Other instructions, from other repositories
agent CLAUDE.md
Instructions for 1mcp-app/agent, covering claude.md, repository information, development commands, setup and build & run.
activepieces CLAUDE.md
Claude Code instructions for activepieces/activepieces, covering activepieces, architecture (non-obvious rules), coding conventions, query error handling and key utilities (@activepieces/shared).
mcp-server-fuzzer AGENTS.md
Instructions for Agent-Hellboy/mcp-server-fuzzer, covering ai agent instructions, branch creation, commit message style, lint and tests and tox venv setup.
raindrop-mcp dxt.instructions.md
Instructions for adeze/raindrop-mcp: I want to build this as an MCP Bundle, abbreviated as "MCPB". Please follow these steps.
mcpscore AGENTS.md
Instructions for mcp-box/mcpscore, covering agent instructions for mcpscore and gotchas (learned the hard way).
mcp-scribe CLAUDE.md
Instructions for The-Swarm-Corporation/mcp-scribe, covering claude.md, what this is, commands, formatting: black only. do not run ruff format and architecture.