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/psycgod/sage/claude-mdgit clone --depth 1 https://github.com/PsYcGoD/sageWrote 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/psycgod/sage/claude-md)<a href="https://agentmods.dev/instructions/psycgod/sage/claude-md"><img src="https://agentmods.dev/badge/instructions/psycgod/sage/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 | $0.00959 | $0.00959 |
| Opus 5 | $0.00479 | $0.00479 |
| Sonnet 5 | $0.00192 | $0.00192 |
| Haiku 4.5 | $0.00096 | $0.00096 |
Grade A, and why
sage 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 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.
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 — 100 lines — stays where its author put it; the contents beside it link to each section on GitHub.
SAGE MANAGED BLOCK START
SAGE Integration - MANDATORY
When a user asks for any coding, repo, file, debugging, documentation, or project task, start through SAGE automatically. The user should not need to type sage.
First actions for normal repo work:
- Route every terminal command through a SAGE wrapper
- Use native file/search/edit tools normally unless the user has manually installed and enabled the optional SAGE MCP server.
Route every shell command through SAGE. Do not run bare terminal commands.
Allowed wrappers:
- PyPI/pip install:
sage run -- <command> - npm/npx install:
npx -y psycgod-sage run -- <command>
Rules:
- Python:
sage run -- python script.pyornpx -y psycgod-sage run -- python script.py - Packages:
sage run -- npm installornpx -y psycgod-sage run -- npm install - Tests:
sage run -- pytestornpx -y psycgod-sage run -- pytest - If the wrapper fails before the target command starts, stop and report the wrapper failure.
Why:
- Tracks local command history for the user
- Compresses noisy output before it reaches AI context
- Captures safe proof metrics for token savings and ML prediction
- Keeps raw commands and output local unless the user enables higher telemetry
Examples:
- Wrong:
python test.py - Correct:
sage run -- python test.py - Correct with npm/npx:
npx -y psycgod-sage run -- python test.py - Wrong:
git status - Correct:
sage run -- git status
Optional MCP:
- SAGE MCP is not auto-enabled because MCP stdio sessions can disconnect in long-running AI-agent sessions.
- If SAGE MCP tools are manually installed and visible, you may use them.
- If SAGE MCP tools are unavailable or disconnected, keep working with native file/search/edit tools and continue wrapping terminal commands through SAGE.
Example user request:
- User: "Please help me with my general book in this folder"
- Correct agent start: inspect/read/edit files with available native tools, then use
sage run -- <command>only when a terminal command is needed. - Wrong agent start: bare
python,git,npm,ls, or other terminal commands without SAGE.
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 · 100 lines · 959 tokens per session scan A 5a9678572b8c
sage CLAUDE.md is an instructions file published in the GitHub repository PsYcGoD/sage (10 stars, last pushed 1mo ago), licensed MIT. It adds 959 tokens to every session, about $0.0048 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
ui-ux-pro-max-skill CLAUDE.md
Claude Code instructions for nextlevelbuilder/ui-ux-pro-max-skill, covering claude.md, project overview, search command, architecture and sync rules.
tweakcc AGENTS.md
AGENTS.md instructions for Piebald-AI/tweakcc, covering agents.md, development commands, build, testing and linting & formatting.
NeoHaskell AGENTS.md
AGENTS.md instructions for neohaskell/NeoHaskell, covering neohaskell — agent guide, hard rule, style (mandatory), commands (pointers, not a duplicate) and change flow (phase 5) — spec-gated, two human touchpoints.
tokf CLAUDE.md
Claude Code instructions for mpecan/tokf, covering tokf — development guidelines, project philosophy, commits, code quality and testing.
vue-tui AGENTS.md
AGENTS.md instructions for vuejs-ai/vue-tui, covering common pitfalls & best practices, project context records (pcr), using vite+, the unified toolchain for the web and review checklist.
axe AGENTS.md
Instructions for jrswab/axe, covering axe — agents.md, what this is, design philosophy, non-obvious constraints and what good contributions look like.