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/bynoe17/dkwtmcp/claude-mdgit clone --depth 1 https://github.com/bynoe17/DkwtMCPWrote 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.
[](https://agentmods.dev/instructions/bynoe17/dkwtmcp/claude-md)<a href="https://agentmods.dev/instructions/bynoe17/dkwtmcp/claude-md"><img src="https://agentmods.dev/badge/instructions/bynoe17/dkwtmcp/claude-md.svg" alt="Measured on agentmods" height="20"></a>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 | $0.02692 | $0.02692 |
| Opus 5 | $0.01346 | $0.01346 |
| Sonnet 5 | $0.00538 | $0.00538 |
| Haiku 4.5 | $0.00269 | $0.00269 |
Grade A, and why
DkwtMCP CLAUDE.md scanned grade A with 0 findings 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 5d 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Instrucciones del proyecto — DkwtMCP
Contexto para cualquier sesión que trabaje en este repo. Léelo antes de tocar código.
Qué es
MCP local (stdio) en Python del equipo DkwtMCP que conecta Claude Desktop con Garmin y Apple Health, lectura + escritura. Uso personal, no un servicio alojado. Paquete Python: dkwt_mcp. Referencia conceptual: FitMCP (https://fitmcp.tech).
Strava se retiró del alcance por ahora (sus términos 2025 prohíben usar sus datos en IA). Si se reincorpora, seguir el patrón de providers.
Reglas de diseño (no negociables)
- Framework FastMCP (
mcp[cli]). La instancia vive enapp.py; los providers la importan y registran tools con@mcp.tool. - Un tool = un workflow útil, no un envoltorio 1:1 de la API. Nombres con prefijo de plataforma:
garmin_*,apple_*, másfitness_*para transversales. - Toda tool: modelo Pydantic v2 de entrada (
extra="forbid"),annotations(readOnlyHint/destructiveHint/idempotentHint/openWorldHint), docstring con Args/Returns, y salida porcommon.as_output(markdown|json). - Errores siempre vía
common.handle_api_error(mensajes accionables). Nuncaexceptsilencioso que oculte el fallo al agente. - I/O de red en async.
garminconnectes síncrono → envolver conasyncio.to_thread. - Secretos solo en
.env(git-ignorado). Nada de credenciales en el código.
Restricciones externas a recordar
- Garmin usa la librería no oficial
python-garminconnect(>=0.3), frágil ante cambios anti-bots de Garmin. Fijar versión; si el login falla, revisar si hay versión nueva.- Auth: el servidor SOLO reutiliza tokens en
~/.dkwtmcp/garmin/. El primer login (con posible 2FA) se hace conscripts/garmin_login.py. El servidor nunca pide input interactivo (colgaría stdio). Si no hay tokens, las tools devuelven un error que indica ejecutar el script.
- Auth: el servidor SOLO reutiliza tokens en
- Apple Health es solo lectura vía export XML → SQLite. No hay escritura en HealthKit.
Estado actual (Fase 3 Garmin escritura)
- Esqueleto renombrado a
dkwt_mcp, servidor FastMCP"DkwtMCP", arranca y registra tools según flags. - Garmin lectura ✅:
garmin_daily_readiness,garmin_get_activities,garmin_get_activity,garmin_sleep,garmin_hrv. Probado con cuenta real (login + get_activities OK). - Garmin escritura ✅:
garmin_schedule_workout(dry_run=True por defecto, seguridad),garmin_list_workouts,garmin_delete_workout(requiere confirm=true). Payload construido congarminconnect.workout(typed); soporta targets de power (W) y heart_rate (ppm). - Transversal ✅:
fitness_planned_vs_actual(cruza programados vs. actividades reales por fecha, calcula adherencia). Parseo defensivo deget_scheduled_workouts(forma de respuesta a confirmar contra cuenta real). - Plan de fuerza semanal ✅ (puente entre proyectos):
fitness_set_week_plan(escribe) +fitness_get_week_plan(lee) sobre~/.dkwtmcp/plan/semana.json. La fuerza se organiza en OTRO proyecto de Claude ("Pirinexus Training 2026-Noe") y cambia cada semana; como los proyectos no comparten contexto, ese proyecto vuelca la semana confitness_set_week_plany el brief la lee. Prioridad del brief para "planificado_hoy": entrenado hoy > Garmin programado (e-bike) > plan de fuerza semanal (archivo local) > calendario fijo del perfil (ultimo recurso). Requiere que el proyecto de planificacion tenga el MCP DkwtMCP conectado (mismo servidor, mismo archivo). - Apple ✅ (para quien lleve Apple Watch):
apple_import_export+apple_query_metric(sueño, HRV, FC reposo, pasos, energia) -> SQLite en ~/.dkwtmcp/apple/. Tambienapple_get_workouts(entrenos del Apple Watch, p.ej. fuerza, desde data.workouts). Dos formatos: export.xml/zip de la app Salud (iterparse) y Health Auto Export JSON (carpeta con .json y/o .zip, importa todo y deduplica; auto-incremental, recomendado). En Health Auto Export, Metricas y Workouts son categorias separadas: hay que exportar ambas. Detecta el formato solo. Activar con ENABLE_APPLE=true.
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.
- 5d ago First seen · 72 lines · 2,692 tokens per session scan A 3863b8d55e2a
DkwtMCP CLAUDE.md is an instructions file published in the GitHub repository bynoe17/DkwtMCP (0 stars, last pushed 14d ago), licensed MIT. It adds 2,692 tokens to every session, about $0.0135 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
codex AGENTS.md
AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
spec-kit AGENTS.md
AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).