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/acogood/diffmode_free/claude-mdgit clone --depth 1 https://github.com/acogood/diffmode_freeWhat 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 | $0.02528 | $0.02528 |
| Opus 5 | $0.01264 | $0.01264 |
| Sonnet 5 | $0.00506 | $0.00506 |
| Haiku 4.5 | $0.00253 | $0.00253 |
Grade B, and why
diffmode_free CLAUDE.md scanned grade B with 1 finding 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 2d 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.
`~/.claude/settings.json` → `extraKnownMarketplaces.diffmode-free.source` points at. A How it starts
The opening of the file, as written. The whole thing — 151 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
Guidance for AI agents working in the Diffmode Growth Tactics standalone repo.
What this repo is
The free Diffmode growth-ideation pipeline, packaged as a portable Claude Code plugin
(+ a Codex scaffold). It takes a founder from a 2-minute intake to a synthesis.md of
7-9 novel demand-gen tactic IDEAS and stops at synthesis. It is an extract of the
private ai-cmo monorepo's pipeline-skills/ — re-pointed so it runs standalone, leaking
none of the paid product.
This is a prompt-based system, not traditional software: no build, no runtime. The
"code" is Markdown (SKILL.md bodies, agent defs, orchestrator commands) executed by a
coding-agent runtime. Outputs are Markdown/JSON files written to a per-run workspace.
Buyer-facing essence / moat framing lives in README.md and plugin/README.md.
The moat (do not erode it)
The synthesis method is what this repo gives away. Paid + proprietary (never ship,
never reference content of): the 576-vector growth database (tactics_DB/ in the private
repo), its intelligence layer + saturation/anti-vector tracking, and the downstream
prioritization + implementation stages. The free pipeline runs the same method on a
deliberately weaker per-run LIGHT DB mined fresh from public case studies, and stops.
Clean-room rule (moat-critical): growth-factors-mining and the whole synthesis chain
MUST NOT read anything under tactics_DB/. Mentions of tactics_DB/ in this repo are only
prohibitions / moat framing — never actual DB content, vector counts (no 542/8,238), or
prompts/ source files. Keep it that way.
Architecture — single source of truth + what ships
- One repo, one physical copy of the skills. The 13 skills live once in
plugin/skills/. Edit them there.codex/.agents/skills/*are relative symlinks (../../../plugin/skills/<name>) — Codex reads the same files; do NOT create a second copy. - Only
plugin/ships to Claude users.marketplace.json(repo root) declaressource: ./plugin, so Claude copies just that dir to its cache.docs/,codex/, and the rootREADME.mddo not ship — put internal notes there, not inplugin/. - Codex reads a checked-out repo (not a copy), so the symlinks resolve. The Codex side is a
second, working orchestrator —
codex/orchestrate.pydrives the same DAG,codex/codex_dispatch.pydispatches workers, andcodex/checks.pyholds the deterministic stage checks.checks.pyis the more precise of the two implementations (per-stage completeness anchors with rationale, substring match semantics, the 15-40 gate vs 20-40 acceptance split): when a structural check needs fixing, port from it rather than reinventing the check in prose instart.md— and change both in the same commit. Seecodex/CODEX.md.
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.
- 2d ago First seen · 151 lines · 2,528 tokens per session scan B 1d10ea1e6023
diffmode_free CLAUDE.md is an instructions file published in the GitHub repository acogood/diffmode_free (160 stars, last pushed 22d ago), licensed Apache-2.0. It adds 2,528 tokens to every session, about $0.0126 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
growth-lab AGENTS.md
Instructions for tsingyuai/growth-lab, covering growth lab agent runtime, 用户问“你能做什么”时, 接入产品工作区, 产品代码就在当前工作区 and 产品代码在相邻或其他本地目录.
newsjack AGENTS.md
Instructions for elvisun/newsjack, covering agents.md, repo shape, canonical boundaries, minimal memory and discovery doctrine.
newsjack CLAUDE.md
Instructions for elvisun/newsjack: Load AGENTS.md before editing this repo — follow it.
ai-marketing-skills AGENTS.md
Instructions for superamped/ai-marketing-skills, covering ai marketing skills for ai agents, available skills, ads, content and conversion.
gtm-brain-builder CLAUDE.md
Instructions for Andytoizer/gtm-brain-builder, covering agent instructions, public packaging rules and validation.
builder-growth CLAUDE.md
Instructions for RBraga01/builder-growth, covering builder-growth — claude configuration, pack version, file conventions, skill trigger sequence and agent usage.