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/metedata/remarc/claude-mdgit clone --depth 1 https://github.com/metedata/RemarcWrote 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/metedata/remarc/claude-md)<a href="https://agentmods.dev/instructions/metedata/remarc/claude-md"><img src="https://agentmods.dev/badge/instructions/metedata/remarc/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 | $0.02084 | $0.02084 |
| Opus 5 | $0.01042 | $0.01042 |
| Sonnet 5 | $0.00417 | $0.00417 |
| Haiku 4.5 | $0.00208 | $0.00208 |
Grade D, and why
Remarc CLAUDE.md scanned grade D with 2 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 4d 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
**Codex is install-only: there is no cleanup, no latch, no durable flag, and it deletes nothing.** The app simply stopped writing `[mcp_servers.remarc]` into `~/.codex/config.toml`. Anyone carrying that table from an old Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
rm -rf app/DerivedData How it starts
The opening of the file, as written. The whole thing — 116 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Remarc
macOS menu bar app (SwiftUI + AppKit) for contextual commenting on text selections.
- Bundle ID:
com.metepolat.Remarc - Min macOS: 14.0 (
MACOSX_DEPLOYMENT_TARGETinapp/Config/Shared.xcconfigis the source of truth), Swift 6.0+, LSUIElement (no dock icon) - Debug log:
/tmp/remarc_debug.log(Debug builds only; release builds write no log file unlessdefaults write com.metepolat.Remarc debugFileLoggingEnabled -bool YES) - Data:
~/Library/Application Support/Remarc/comments.json(legacy fallback:data.json)
Agent integrations
Three harnesses, two different delivery models. Do not assume they work the same way.
Claude Code and Codex: plugin-based. Both install from the marketplace at https://github.com/metedata/remarc-agent-plugins, which is not built into the app. Two plugins:
remarc@remarc- required. MCP server + skill.remarc-hooks@remarc- optional, experimental. Session-lifecycle hooks (off by default; user must/plugin install remarc-hooks@remarcexplicitly).
Cursor: still fully app-managed. The app writes ~/.cursor/mcp.json and ~/.cursor/skills/remarc/SKILL.md via HarnessIntegrationManager / CursorMCPInstaller / SkillInstaller, driven by the Enable toggle in Preferences. Cursor's repo marketplaces are Teams/Enterprise-only, so there is no plugin path for individual users. Deliberately deferred, not overlooked - see docs/superpowers/plans/2026-08-06-codex-cursor-plugin-migration.md.
Per-harness plugin manifests (important)
The plugin repo ships THREE manifests over one shared payload, because the harnesses disagree about paths:
.claude-plugin/plugin.json+ root.mcp.jsonusing${CLAUDE_PLUGIN_ROOT}. Claude Code substitutes it..codex-plugin/plugin.json->codex-mcp.jsonusing"args": ["mcp/dist/index.js"], "cwd": ".". Codex does NOT substitute${CLAUDE_PLUGIN_ROOT}(nor${PLUGIN_ROOT}/${CODEX_PLUGIN_ROOT}) in plugin MCP args, and sets no plugin-root env var - it passes the literal string through, so the server never starts. A relativecwd, resolved against the installed plugin root, is the documented replacement (openai/codex#19582, discussion #28145). Codex prefers.codex-pluginover.claude-pluginwhen both exist, which is what keeps Claude Code unaffected..cursor-plugin/plugin.json- groundwork for a future Cursor Marketplace listing only. No app code consumes it.
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.
- 4d ago First seen · 116 lines · 2,084 tokens per session scan D ef16f15006c9
Remarc CLAUDE.md is an instructions file published in the GitHub repository metedata/Remarc (66 stars, last pushed 13d ago), licensed MIT. It adds 2,084 tokens to every session, about $0.0104 per session on Opus 5. A static security scan graded it D with 2 findings (reads agent configuration directories, recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
Peekaboo AGENTS.md
AGENTS.md instructions for openclaw/Peekaboo, covering repository guidelines, start here, project structure & modules, build, test, and development commands and coding style & naming conventions.
Applite CLAUDE.md
Instructions for milanvarady/Applite, covering claude.md, project overview, build & run, architecture and data flow.
palmier-pro AGENTS.md
AGENTS.md instructions for palmier-io/palmier-pro, covering palmierpro, build, agents, engineering approach and code style.
scarf AGENTS.md
Instructions for awizemann/scarf: This repo carries its own agent memory — plain files in git, served by the in-repo memophant-mcp MCP server, managed by the Memophant macOS app. It works the same for every agent (Claude Code, Codex, Cursor, Gemini, Copilot). This block is regenerated between the memophant markers…
scarf copilot-instructions.md
Instructions for awizemann/scarf: File memory notes under one of six folders (architecture/conventions/decisions/operations/project/roadmap), never the root. When a note is grounded in code, pass sourcepaths (the repo files it depends on) so Memory Health can drift-check it — an unanchored code note can't be kept…
mcp-server-apple-events CLAUDE.md
Claude Code instructions for FradSer/mcp-server-apple-events, covering claude.md, commands, run all tests, run a single test file and run tests matching a pattern.