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/netixc/stremio-mcp/agents-mdgit clone --depth 1 https://github.com/netixc/stremio-mcpWrote 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/netixc/stremio-mcp/agents-md)<a href="https://agentmods.dev/instructions/netixc/stremio-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/netixc/stremio-mcp/agents-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 | $0.01406 | $0.01406 |
| Opus 5 | $0.00703 | $0.00703 |
| Sonnet 5 | $0.00281 | $0.00281 |
| Haiku 4.5 | $0.00141 | $0.00141 |
Grade A, and why
stremio-mcp AGENTS.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 — 51 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Project Instructions
Authoritative commands
Run commands from the repository root with Python 3.10+ and uv installed.
- Locked setup:
uv sync --locked. This creates or updates.venvfromuv.lock; it may download packages but does not use TMDB, Stremio, or ADB credentials. - Unit tests:
uv run --locked python -m unittest discover -s tests -v. - Source compilation:
uv run --locked python -m compileall -q src tests. - Package build:
uv build. This creates ignored artifacts underdist/. - Full CI-equivalent verification: run locked setup, unit tests, source compilation, and package build in that order. CI runs the first three checks on Python 3.10 through 3.14 and builds on 3.12.
- These checks provide no linting, type checking, or coverage gate. Existing unit tests mock dispatch boundaries and do not contact TMDB, Stremio, or an Android device.
Change-to-check mapping
- Changes to
src/stremio_mcp.pyortests/→ run unit tests and source compilation. - Changes to
pyproject.tomloruv.lock→ runuv sync --locked, unit tests, anduv build. - Packaging or Python-version compatibility changes → run the full CI-equivalent verification; consult
.github/workflows/ci.ymlfor the supported Python matrix.
Project-specific constraints
pyproject.tomlanduv.lockare the install and CI dependency sources. Keep them synchronized when dependencies change.- Runtime configuration is read from environment variables when
src/stremio_mcp.pyis imported.TMDB_API_KEYenables network search,STREMIO_AUTH_KEYenables credentialed library access, andANDROID_TV_HOSTenables commands to a physical Android TV. - Do not run live MCP calls or ADB commands with real configuration as routine verification: search and library reads contact external services; library add/remove mutates the user's Stremio account; playback, navigation, volume, and power mutate a physical device.
- All outbound HTTP goes through the single
AsyncHTTPClientinsrc/stremio_mcp.py. Never add a synchronous HTTP call or a per-call client: synchronous I/O in an async MCP handler blocks the event loop and freezes unrelated device controls, and a second client escapes the configured timeout, response-size, and pool bounds. Every request needs explicit connect/read/write/pool timeouts. - Credentials must never reach a log record, a traceback, or a returned MCP error. Describe network failures with the category/host/status that
HTTPClientError.summary()produces; never log a prepared URL, a request payload, or a raw upstream exception.redact_secrets()andSecretRedactingFilterare the backstop, not the primary defence. Prove any new failure path with a sentinel-secret test. - Library mutations require an explicit IMDb ID and content type, use
_idfor Stremio datastore identity, preserve watch state on re-add/remove, and verify each write with a follow-up read. Cover these boundaries with mocks; never use a real account for routine tests. - Library reads return typed outcomes (
LibraryRead/LibraryListRead/MetaRead) that separate an authoritative not-found from an error. Mutations must fail closed: abort without writing on any read error,_idmismatch, duplicate row, unrequested extra row, or type mismatch. Never infer "absent" from a failed read. - Native
adbis the transport boundary. Modern Android Wireless Debugging uses TLS (STLS), which pure-Python ADB clients (including the formeradb-shelldependency) do not support; do not replace native ADB with a client that lacks this protocol or that cannot run shell diagnostics for playback status. - Wireless Debugging exposes separate, often ephemeral pairing and connection ports. Never assume the pairing port is the runtime port or that modern devices use legacy port
5555. Official wireless debugging on TV requires Android 13+; hosts should use a current Platform Tools release (minimum wireless-era 30.0.0+, prefer latest stable). - On macOS, Local Network permission applies to the
adbbinary. The supported pattern is for a permitted GUI terminal to start the shared ADB server, while MCP and other tools use it as localhost clients; automated tooling must not runadb kill-serveroradb start-serveror otherwise manage that server lifecycle. - A series deep link requires both season and episode; movies and series use different Stremio URI forms. Preserve this distinction and cover dispatch or URI changes with mocked tests.
- Playback parsing must remain scoped to Stremio's media-session block because other Android sessions can overwrite state. Preserve support for numeric and named states, monotonic position extrapolation, and extractor-based duration fallback in mocked tests.
- Claimed media-session
PLAYINGmust be corroborated with a started Stremio-ownerAudioTrackbefore reporting healthy playback; otherwise demote tostalledand do not extrapolate position. Stremio often freezes rawposition/updatedeven during real play, so dual-sampling the session alone cannot prove liveness. media_stopsuccess is a post-condition (no active playback), not ADB acceptingKEYCODE_MEDIA_STOP. Stremio/VLC commonly ignores STOP while accepting pause/play; keep the bounded verify → pause+back →am force-stop com.stremio.onepath and fail closed if the session still plays.- Live TV dogfood: after starting a stream, wait for buffering/settle before treating
playback_statusor pause/resume/stop samples as authoritative (early samples often look stalled/unknown at position 0). Never press select/enter on the launcher — that can open an unrelated app; only center whilecom.stremio.oneis focused. dist/,.venv/, and Python cache files are generated outputs; do not edit them directly or include them in source 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.
- 5d ago First seen · 51 lines · 1,406 tokens per session scan A 18a788ec0270
stremio-mcp AGENTS.md is an instructions file published in the GitHub repository netixc/stremio-mcp (6 stars, last pushed 7d ago), licensed MIT. It adds 1,406 tokens to every session, about $0.0070 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
openrouter-mcp-multimodal AGENTS.md
AGENTS.md instructions for stabgan/openrouter-mcp-multimodal, covering agent instructions, before you ship, releasing (read this before publishing), short version and version files (must all match package.json).
cyxj-groksearch CLAUDE.md
Claude Code instructions for chenyuxiaojin/cyxj-groksearch, covering claude.md, 常用命令 and 结构与坑.
Plonk AGENTS.md
AGENTS.md instructions for ostapondo/Plonk, covering agent rules, layout, adding a module, build & verify and code style.
mcp-dockhand copilot-review.instructions.md
Copilot instructions for strausmann/mcp-dockhand: When reviewing pull requests in this repository.
minecraft-dev-mcp AGENTS.md
Instructions for MCDxAI/minecraft-dev-mcp, covering agent handbook – minecraft dev mcp, project snapshot, what agents should prioritize, architecture wayfinder (src/) and available mcp tools (for llm surfaces).
bray AGENTS.md
AGENTS.md instructions for forgesworn/bray, covering agents.md — nostr-bray, what this is, build & test, architecture and tool groups.