What is skope CLAUDE.md?

Repository instructions for skope, a local-first tool that finds news relevant to a person’s real-world situation and watches for gaps in their usual sources. It uses an AI assistant for searching and writing summaries, while skope stores profiles, duplicate checks, scores, and feedback.

Use them when developing skope’s news collection, relevance scoring, deduplication, profile, feedback, MCP tools, or command-line interface. MCP lets an AI assistant call the project’s tools.

They define the project’s idea of relevance and separate AI judgment from the parts that must behave consistently. This helps prevent random news, duplicate items, and violations of the project’s architecture.

skope CLAUDE.md — Instructions file for coding agents by evan-moon: Repository instructions for skope, a local-first tool that finds…

Install

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.

agentmods

npx agentmods add instructions/evan-moon/skope/claude-md

Clone the repo

git clone --depth 1 https://github.com/evan-moon/skope

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/evan-moon/skope/claude-md.svg)](https://agentmods.dev/instructions/evan-moon/skope/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/evan-moon/skope/claude-md"><img src="https://agentmods.dev/badge/instructions/evan-moon/skope/claude-md.svg" alt="Measured on agentmods" height="20"></a>

Per session 1,371 This file is loaded in full into every session.

When invoked 1,371 The same file — it is already loaded in full.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

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	$0.01371	$0.01371
Opus 5	$0.00685	$0.00685
Sonnet 5	$0.00274	$0.00274
Haiku 4.5	$0.00137	$0.00137

Measured 5d ago against content hash 25bcf8eeab2f, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

skope 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.

CLAUDE.md · 88 lines

How it starts

The opening of the file, as written. The whole thing — 88 lines — stays where its author put it; the contents beside it link to each section on GitHub.

skope — Project Context

skope is a local-first personalized news intelligence tool — a lens + watcher on the world. Not "the world's news" but "news that reaches you". It scans broadly, aligns to your situation (Reachability), and watches you against your own echo chamber. Target audience: developers worldwide (Hacker News / GeekNews demographic), any MCP client. Herald ecosystem family member, alongside [firma] (assets) and [memex] (memory).

Core philosophy

Lens, not filter. Relevance = Reachability (does a causal path reach you?), not "is this my interest?". Broadens without becoming random noise. brazil rate hike reaches you via currency exposure → relevant; a Brazilian's daily life has no path → excluded.
Collection is active (non-deterministic), the ledger is deterministic. The LLM (the MCP caller / orchestrator) decides what to search, synthesizes causal narrative, infers interests. skope owns the deterministic ledger: profile, dedup (URL hash), rule-based impact scoring, Effective-N concentration, the read/mute feedback log, and the rule-match seed of reachability.
MCP-first, CLI is the safety net. Analysis/synthesis is MCP tools. The CLI is for setup, manual profile editing, and read-only inspection.
Federation, not dependency. skope owns its interest profile (single source of truth). firma / memex are optional adapters — the LLM reads them and feeds skope via update_profile. skope never imports another tool; it does not break if firma dies.

Architecture

Turborepo monorepo with Yarn Berry (nodeLinker: node-modules):

apps/
  cli/                     — @evan-moon/skope  TypeScript CLI (Commander + Clack). Safety net only.
  mcp/                     — @skope/mcp         MCP server (shares ~/.skope/skope.db with CLI)
packages/
  domain/                  — Ports (interfaces + types), no external-API dep
    domain/                — @skope/domain — Profile/Axis, Article/Tier, ImpactScore/ReachabilitySeed,
                             Brief/Concentration, SearchProvider port
  external-api/            — Raw API clients, no @skope/domain dep
    tavily/                — @skope/tavily — Tavily Search + extract client
    source-trust/          — @skope/source-trust — Tier-1 global anchors + country→Tier-2 seed data
  adapter/                 — The only layer that imports both sides
    adapters/              — @skope/adapters — SearchProvider registry (tavily + source-trust)
  use-case/                — Business logic (depends on domain only)
    profile/               — @skope/profile  — axis upsert, weight normalization (sum=1.0), 6-axis cap, cold-start
    discovery/             — @skope/discovery — scan (query budget, incremental) + score (dedup, rule impact, reachability seed)
    brief/                 — @skope/brief    — 2-layer assembly: [your radar] + [world headlines] + concentration
    watch/                 — @skope/watch    — Effective-N (Inverse Simpson) concentration watcher
  shared/                  — Cross-cutting
    db/                    — @skope/db    — Drizzle schema + repository (~/.skope/skope.db)
    utils/                 — @skope/utils — pure helpers (url hash, Effective-N math, date)

Read the full file on GitHub · 88 lines

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 · 88 lines · 1,371 tokens per session scan A 25bcf8eeab2f

Subscribe to this mod's changes

skope CLAUDE.md is an instructions file published in the GitHub repository evan-moon/skope (0 stars, last pushed 1mo ago), licensed MIT. It adds 1,371 tokens to every session, about $0.0069 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.

ai anthropic claude cli local-first mcp

skope CLAUDE.md

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade A, and why

Nothing flagged

How it starts

skope — Project Context

Core philosophy

Architecture

What this file has done since we first saw it

Other instructions, from other repositories

gini-agent AGENTS.md

plur CLAUDE.md

barkeep AGENTS.md

rondoflow CLAUDE.md

blz AGENTS.md

linggen CLAUDE.md