Give agents reverse engineering output they can check, not guess

What you end up with

Three design ideas transfer to any MCP server: cap and page every list, offer a 'since' call so the agent only reads changes, and label results as checked or unchecked.

What you get

• A plain-language summary of the workflow
• A fill-in prompt template
• What to watch out for
• A link to the original source

Who it is for

Builders who already use AI agents and want a proven workflow they can copy into their own project.

Works with

Solid chips: the agent is named in this listing's own tools list. Dashed chip: Plain Markdown, so it also pastes into other agents such as Grok Bot, Hermes, OpenClaw or Muse. We list those as compatible by format only, not as tested.

Tools needed

Cost to run: Free with the agent plan you already use.
Time to result: One working session

How it works

1. Install the binary and connect it as an MCP server

   Install the binary and connect it as an MCP server

   Expected: Done before you move to the next step.

2. Ask for the functions that touch a given string or import, paged

   Ask for the functions that touch a given string or import, paged

   Expected: Done before you move to the next step.

3. Read decompiled C and note which blocks are marked checked

   Read decompiled C and note which blocks are marked checked

   Expected: Done before you move to the next step.

4. Treat unchecked blocks as leads to confirm, not facts

   Treat unchecked blocks as leads to confirm, not facts

   Expected: Done before you move to the next step.

5. Write what you learned into notes the next session can load

   Write what you learned into notes the next session can load

   Expected: Done before you move to the next step.

Sample prompt from the pack

Apply this workflow to [your project]:
1. Install the binary and connect it as an MCP server
2. Ask for the functions that touch a given string or import, paged
3. Read decompiled C and note which blocks are marked checked
4. Treat unchecked blocks as leads to confirm, not facts
5. Write what you learned into notes the next session can load
Stop and ask me before anything irreversible: deploys, payments, sends or deletes.

Free preview of the file

# Give agents reverse engineering output they can check, not guess

PaperChase agent use case (free feed entry), version 1.0. Source: [@dear_prism on X (GitHub: imjustprism)](https://x.com/dear_prism/status/2114486515808219316), 2026-10-25.

## What they did

The author argues existing disassemblers were made for a person at a screen, with output sized for eyes rather than a context window. borer exposes about 41 tools that the author puts at roughly 7,000 tokens of definitions. List results are paged and capped, follow-up calls can ask only for what changed since a point, and every block of decompiled C is run in an emulator from seeded states and compared with the original instructions. Anything it cannot model stays marked as raw assembly instead of being guessed, and the output says how much was checked.

## Worth knowing

Three design ideas transfer to any MCP server: cap and page every list, offer a 'since' call so the agent only reads changes, and label results as checked or unchecked. The project warns it is early and tool names may change. Use it only on binaries you have the right to analyze.

Worth knowing

• Three design ideas transfer to any MCP server: cap and page every list, offer a 'since' call so the agent only reads changes, and label results as checked or unchecked. The project warns it is early and tool names may change. Use it only on binaries you have the right to analyze.
• Figures are as reported by the original author and were not re-measured by us.

Details

Version 1.0 · Updated Oct 27, 2026 · Markdown version of this page
License: Personal and commercial use. Credit the original source when you share the workflow. Download terms

Related

Prompt pack$14

AGENTS.md, CLAUDE.md and Cursor rules starter files

by PaperChaseWebb

Six fill-in instruction files for coding agents (base, Next.js plus Supabase, FastAPI, browser game, CLAUDE.md pointer, scoped Cursor rules) and 8 prompts to write, trim and test your own.

Works withClaude CodeCodexCursorPlain .md

Vibe coding · Beginner · 20 min setup

Prompt pack$19

Ask-first rules for AI agents: hard nos, approval cards and a test set

by PaperChaseWebb

Paste-in rules that make an agent ask before it sends, spends, posts or deletes, define what a real yes is, and stop retries after a no, plus 12 test scenarios to check it.

Works withClaude CodeCodexClaudeCursor+2Plain .md

Solo business · Beginner · 20 min setup