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/gerardordz96/deepfork/claude-mdgit clone --depth 1 https://github.com/GerardoRdz96/deepforkWrote 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/gerardordz96/deepfork/claude-md)<a href="https://agentmods.dev/instructions/gerardordz96/deepfork/claude-md"><img src="https://agentmods.dev/badge/instructions/gerardordz96/deepfork/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 | $0.01083 | $0.01083 |
| Opus 5 | $0.00541 | $0.00541 |
| Sonnet 5 | $0.00217 | $0.00217 |
| Haiku 4.5 | $0.00108 | $0.00108 |
Grade A, and why
deepfork 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 — 45 lines — stays where its author put it; the contents beside it link to each section on GitHub.
DeepFork — project context for Claude Code
DeepFork is an agent skill with one idea: don't fork the code, fork the design. Point it at any open-source repo and it produces a clean explanation of how the repo actually works, a rebuildable behavioral blueprint, and a clean-room reimplementation in your stack with your changes. Public OSS, MIT, repo GerardoRdz96/deepfork. Powered by graphify.
This file is authoritative project context. Verify claims against the code before acting; do not improvise facts about this repo.
What this repo is
DeepFork is not an application — it is a Claude Code skill distributed as a repo (works with any agent that reads skills). The whole product is the skill prompt plus its examples. There is no build, no package, no test suite to run here — the "execution" is an agent following skills/deepfork/SKILL.md against a target repo.
Repo layout (small by design — 7 tracked files):
skills/deepfork/SKILL.md— the skill. The entire pipeline lives here. This is the file you edit to change DeepFork's behavior.examples/— a worked run (micrograd).assets/— the banner.README.md— the public pitch.LICENSE— MIT.deepfork-out/— sample output artifact (the skill writes here at runtime; keep it out of any rebuilt repo).
The pipeline (what SKILL.md encodes — keep these invariants)
A license-gated reverse-engineering pipeline. Every phase's product is a file the user keeps; all outputs go to deepfork-out/<target-name>/.
- Phase 0 — License gate (non-negotiable, never skip). Check the target's license (
gh api repos/<owner>/<repo> --jq .license.spdx_idor the LICENSE file). Proceed only for OSI-approved licenses. No license = all-rights-reserved → stop (an understanding report from public code is still OK; no rebuild). Copyleft (GPL/AGPL) → warn that a closely-derived rebuild may carry obligations; recommend the rebuild also be open source. Record the verdict inLICENSE-GATE.md. - Phase 1 — Acquire.
git clone --depth 1 <target-url> deepfork-out/<target>/source. - Phase 2 — Comprehend (the graph pass). Prefer
graphify(graphify .is free/local tree-sitter AST;graphify cluster-only,graphify . --wiki). Fallback if absent: map entry points, manifest, dir tree, top-5 most-imported modules by hand. Then writeUNDERSTANDING.md— load-bearing pieces (graphify's "god nodes"), data flow, the core trick, surprising couplings. - Phase 3 — Interrogate. Answer a rebuilder's questions via
graphify query / path / explain(or direct reading); append toUNDERSTANDING.md§ Interrogation. Read real code for anything still[INFERRED]the rebuild depends on. - Phase 4 — Blueprint. Write
BLUEPRINT.md— a spec someone could build from without ever seeing the original source. - Phase 5 — Rebuild (clean-room). Build from the blueprint, tests first for the core mechanisms, milestone by milestone. Ship with
ATTRIBUTION.md("Design informed by reverse-engineering <original> (<license>). Implementation is original, built clean-room from a behavioral blueprint.").
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 · 45 lines · 1,083 tokens per session scan A 98e9445ca1aa
deepfork CLAUDE.md is an instructions file published in the GitHub repository GerardoRdz96/deepfork (10 stars, last pushed 2mo ago), licensed MIT. It adds 1,083 tokens to every session, about $0.0054 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
ida-pro-mcp CLAUDE.md
Instructions for mrexodia/ida-pro-mcp, covering claude.md, what this project is, core implementation rules, ida thread safety and api conventions.
m_flow AGENTS.md
AGENTS.md instructions for FlowElement-xinliuyuansu/m_flow, covering m-flow — developer & agent reference, 1. repository map, extension points, 2. local development and python backend (requires python 3.10 – 3.13).
llm-wiki-agent GEMINI.md
Gemini CLI instructions for SamurAIGPT/llm-wiki-agent, covering llm wiki agent — schema & workflow instructions, how to use, directory layout, page format and ingest workflow.
sm64ds-decomp AGENTS.md
AGENTS.md instructions for tangosdev/sm64ds-decomp, covering posting a pr (for ai agents and their humans), the one rule that matters, what a change looks like, which directory under src/ and shared headers (include/).
ontology-atlas AGENTS.md
AGENTS.md instructions for wlsdks/ontology-atlas, covering agents.md — ontology-atlas, product and non-negotiable architecture, start here, structure and routes and operating gates and skills.
ChatCrystal AGENTS.md
AGENTS.md instructions for ZengLiangYi/ChatCrystal, covering agents.md, project overview, commands, development (server port 3721 + client port 13721) and build for production.