/solve - Full Solve Workflow

Autonomously attempt to solve a CTF challenge through the complete workflow.

Usage

/solve <challenge_path_or_url>
/solve <challenge_directory>

Instructions

When the user invokes /solve, execute the full CTF solving workflow:

Phase 1: RECON

1. •Identify all challenge files
2. •Read any README or description
3. •Determine challenge category
4. •Note flag format if specified

Phase 2: ANALYSIS

Based on category, perform deep analysis:

For PWN:

• •Run checksec
• •Identify vulnerability class
• •Find target functions/addresses
• •Map out exploitation path

For Crypto:

• •Identify cipher/algorithm
• •Find weaknesses
• •Gather parameters (n, e, c for RSA, etc.)

For Web:

• •Map attack surface
• •Identify injection points
• •Check authentication/session handling

For Forensics:

• •Extract hidden data
• •Analyze file structure
• •Check for steganography

For RE:

• •Static analysis of key functions
• •Identify algorithm/logic
• •Find key validation routine

Phase 3: EXPLOIT

1. •

   Select appropriate agent if complex:

   • •pwn-expert for binary exploitation
   • •crypto-solver for cryptography
   • •web-hacker for web challenges
   • •forensics-analyst for forensics
   • •reverse-engineer for reverse engineering
2. •

   Generate exploit using Python tools in tools/

3. •

   Create and test exploit script

Phase 4: VERIFY

1. •Execute exploit
2. •Extract flag
3. •Validate flag format
4. •Report success or iterate

Phase 5: ITERATE (if needed)

If initial approach fails:

1. •Re-analyze with new information
2. •Try alternative techniques
3. •Adjust exploit parameters
4. •Maximum 3 iteration attempts

Phase 6: WRITEUP

On success, offer to generate writeup:

/writeup <challenge_name>

Autonomous Decision Making

During /solve, make intelligent decisions:

1. •Tool Selection: Choose appropriate tools based on challenge type
2. •Agent Delegation: Use specialized agents for complex sub-tasks
3. •Error Recovery: Adapt approach based on failures
4. •Resource Management: Don't spend excessive time on dead ends

Output Format

Provide status updates at each phase:

## Solving: <challenge_name>

### Phase 1: RECON ✓
- Found: binary, source code
- Category: PWN
- Flag format: flag{...}

### Phase 2: ANALYSIS ✓
- Vulnerability: Buffer overflow in read_input()
- Protection: NX enabled, no PIE
- Approach: ret2libc

### Phase 3: EXPLOIT
- Generating exploit...
- Testing locally...
- [status updates]

### Phase 4: VERIFY
- Flag captured: flag{...}

### Result: SUCCESS ✓

Failure Handling

If solve fails after attempts:

1. •Document what was tried
2. •Explain blockers encountered
3. •Suggest manual investigation points
4. •Offer partial solution/analysis