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/feanorscodesl/imladris/claude-mdgit clone --depth 1 https://github.com/FeanorsCodeSL/imladrisWrote 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/feanorscodesl/imladris/claude-md)<a href="https://agentmods.dev/instructions/feanorscodesl/imladris/claude-md"><img src="https://agentmods.dev/badge/instructions/feanorscodesl/imladris/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 5.1 | $0.01478 | $0.01478 |
| Opus 5 | $0.00739 | $0.00739 |
| Sonnet 5 | $0.00296 | $0.00296 |
| Haiku 4.5 | $0.00148 | $0.00148 |
Grade A, and why
imladris 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 — 108 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
The design and current contracts live in docs/architecture/ (overview,
pipeline-workflow, configuration-reference, security, deployment, dependencies).
Stack: Python, stdio-only, like thorondor. Python 3.13 +
fastmcp, run as an MCP stdio subprocess.
Name: Imladris. The slug
imladrisis the MCP tool id and server id (FastMCP("imladris")), the console scripts (imladris,imladris-mcp), and the skill name. The user-facing command/prompt is/council. Treatimladrisas a single token; keep it stable.
What this is
A Python MCP server replicating OpenRouter's Fusion: fan a prompt out to a
configurable panel of OpenAI-compatible LLM providers, run an API-side analysis
judge (consensus / contradictions / partial coverage / unique insights / blind
spots), then let the harness's own native model author the final answer from the
analysis plus all raw panel answers. Portable across
Claude Code, Codex, OpenCode, and any MCP-aware harness. Installed via the
one-liner installer and configured via the full-screen imladris TUI.
Golden rules (these shape the whole design — internalise them)
- The harness's native model is always the FINAL AUTHOR. An MCP server cannot
reliably call back into it (MCP
samplingis unimplemented in CC/Codex). So the analysis judge runs API-side and finishes before the tool returns; the native model authors only after, in the outer loop. - The judge analyses; it never merges or authors. Two API-side roles (panel →
analysis judge), then the native author writes from
analysis+ all raw answers. (The old curator pass was removed — D13; anonymization too — D14.) - The engine is stateless for one-shot
/council(derived fromprompt+context) apart from a per-request depth guard. Sessions are the one stateful component:/council-sessionowns a local on-prem store (D15). - Secrets come from env vars referenced by name (
api_key_env), stored in~/.imladris/.env(0600) and loaded at startup. Never hardcode keys, never put them in config files, never log them, never return them fromimladris_status. - Every provider is OpenAI-compatible (
/v1/chat/completions) — OpenRouter, DeepSeek, MiniMax, local vLLM, OpenAI, Groq, Together, … There is no special case. (The old "Anthropic is the only special case" rule is retired; the only Anthropic model in the loop is the harness's native Claude, the final author.) - Partial panel results are normal. A provider failure is a recorded error in
the result, never a raised exception that fails the batch. If
ok == 0, short-circuit to a clean error so the host can fall back. - Judge JSON parsing must degrade gracefully — strip fences/preamble, and on
parse/schema failure record
meta.judge_errorand returnanalysis: null; raw panel answers are returned unconditionally regardless, so the host can still author. Never crash. - stdio only — no Docker, no REST, no exposed port, no bearer token.
- Commit policy: commit/stage only when the user explicitly asks; never add AI tools as authors or co-authors.
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 · 108 lines · 1,478 tokens per session scan A bcd072889415
imladris CLAUDE.md is an instructions file published in the GitHub repository FeanorsCodeSL/imladris (4 stars, last pushed 2mo ago), licensed MIT. It adds 1,478 tokens to every session, about $0.0074 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
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.
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.
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).