v3.5.1: live findings + /finding + Ctrl+O/expand + 3-way /stop (soft validate) + report URL + structured Typst + IIS/CMS/CVE agents

REPL interactivity & findings:
- Live findings registered during a run: /results shows them accumulating;
  /finding opens a selection menu with FULL details (PoC, command, evidence,
  CVSS, OWASP/CWE, remediation). Past runs too.
- /expand (and Ctrl+O) dump the last full, untruncated commands.
- Findings colored by severity in the feed (not all-yellow); confirmed vote = green.

Stop & report:
- CRITICAL: /stop no longer kills validation. New SOFT stop (pool.soft) halts
  launching new agents but lets in-flight + VALIDATION finish — so confirmed
  findings are kept. /stop now asks 3 ways: [1] validate then report,
  [2] report raw (no validation), [3] discard.
- Report file:// URL printed on completion/stop.

Report:
- Typst report restructured: executive summary, a Vulnerability Summary TABLE
  (#, vuln, severity, CVSS, OWASP/CWE), and per-finding sections with criticality,
  CVSS, OWASP/CWE, description/impact, PoC, evidence, remediation. owasp passed through.

Agents: +14 app-stack/CVE (IIS tilde/WebDAV/ViewState/debug/handler-bypass,
CMS fingerprint + WordPress/Joomla/Drupal/default-admin, app-server consoles,
exposed VCS, known-CVE & outdated-component exploitation) → 343 total.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

.gitignore

@@ -102,3 +102,5 @@ neurosploit-rs/runs/
 v34_gui.png
 data/repl_runs.json
 data/repl_history.txt
+.neurosploit/
+/tmp/*

agents_md/vulns/appserver_exposure.md

@@ -0,0 +1,36 @@
+# App-Server Console Exposure Agent
+
+## User Prompt
+You are testing **{target}** for exposed Tomcat/JBoss/Jenkins/Actuator consoles.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Discover
+- Probe `/manager/html`, `/jmx-console`, `/jenkins`, `/actuator`, `/console`, `/admin`
+
+### 2. Assess
+- Test default/weak creds (in scope); check unauth-exposed management endpoints
+
+### 3. Confirm
+- Demonstrate a management action / deploy / info-leak proving exposure (→ often RCE)
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: App-Server Console Exposure at [endpoint]
+- Severity: High
+- CWE: CWE-1188
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Remote code execution / takeover
+- Remediation: Authenticate & network-restrict consoles; remove defaults
+```
+
+## System Prompt
+You are a specialist in exposed Tomcat/JBoss/Jenkins/Actuator consoles. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/aspnet_debug_trace.md

@@ -0,0 +1,36 @@
+# ASP.NET Debug/Trace Exposure Agent
+
+## User Prompt
+You are testing **{target}** for debug/trace enabled in production ASP.NET.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Probe
+- Request `trace.axd`; send `DEBUG` verb; check `<compilation debug=...>` leakage via errors
+
+### 2. Assess
+- Harvest request/session data, stack traces, app internals from trace output
+
+### 3. Confirm
+- Show sensitive runtime data exposed
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: ASP.NET Debug/Trace Exposure at [endpoint]
+- Severity: Medium
+- CWE: CWE-489
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Information disclosure
+- Remediation: Disable debug/trace; custom errors
+```
+
+## System Prompt
+You are a specialist in debug/trace enabled in production ASP.NET. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/aspnet_viewstate.md

@@ -0,0 +1,36 @@
+# ASP.NET ViewState Deserialization Agent
+
+## User Prompt
+You are testing **{target}** for unprotected/known-key __VIEWSTATE deserialization.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Inspect
+- Capture __VIEWSTATE; check if MAC is disabled (enableViewStateMac=false) or a known/leaked machineKey is in play
+
+### 2. Weaponize
+- With a known/guessed machineKey, craft a ysoserial.net ViewState gadget
+
+### 3. Confirm
+- Prove code execution via OOB callback or command output tied to a unique marker
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: ASP.NET ViewState Deserialization at [endpoint]
+- Severity: Critical
+- CWE: CWE-502
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Remote code execution
+- Remediation: Enable ViewState MAC; rotate machineKey; patch
+```
+
+## System Prompt
+You are a specialist in unprotected/known-key __VIEWSTATE deserialization. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/cms_default_admin.md

@@ -0,0 +1,36 @@
+# CMS Admin Panel & Default Creds Agent
+
+## User Prompt
+You are testing **{target}** for exposed CMS admin with weak/default credentials.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Locate
+- Find admin (`/wp-admin`, `/administrator`, `/user/login`, `/admin`)
+
+### 2. Test (in scope)
+- Try supplied/default credentials; respect lockout/ROE — no out-of-scope brute force
+
+### 3. Confirm
+- Show authenticated admin access
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: CMS Admin Panel & Default Creds at [endpoint]
+- Severity: High
+- CWE: CWE-1392
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Full CMS compromise
+- Remediation: Remove defaults; strong creds + MFA; restrict admin
+```
+
+## System Prompt
+You are a specialist in exposed CMS admin with weak/default credentials. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/cms_fingerprint.md

@@ -0,0 +1,37 @@
+# CMS Fingerprint & Version Agent
+
+## User Prompt
+You are testing **{target}** for CMS identification and version disclosure.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Identify
+- Detect CMS via meta generator, paths (`/wp-`, `/sites/`, `/administrator/`), headers, favicon hash
+- Run whatweb/wpscan-style detection without auth
+
+### 2. Version
+- Pin exact version from readme/changelog/asset hashes
+
+### 3. Map
+- List plugins/themes/modules and their versions for CVE correlation
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: CMS Fingerprint & Version at [endpoint]
+- Severity: Info
+- CWE: CWE-200
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Targeted exploitation surface
+- Remediation: Hide version/generator; keep components updated
+```
+
+## System Prompt
+You are a specialist in CMS identification and version disclosure. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/cve_known_exploitation.md

@@ -0,0 +1,40 @@
+# Known-CVE Exploitation Specialist Agent
+
+## User Prompt
+You are testing **{target}** for exploiting known CVEs for the detected stack.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Identify versions
+- From recon, list each component + exact version (server, framework, CMS, plugins, libs)
+
+### 2. Map to CVEs
+- Match versions to known CVEs; prioritise unauth RCE/SQLi/auth-bypass; note CVE id + CVSS
+- Prefer issues with a reliable, non-destructive PoC
+
+### 3. Reproduce safely
+- Run a benign PoC (e.g. a version/echo check or OOB callback) to confirm the CVE is actually present and exploitable — never a destructive payload
+
+### 4. Confirm
+- Report the CVE only when the PoC produced concrete proof (output/OOB); otherwise report it as 'potentially vulnerable (version match, unconfirmed)'
+
+### 5. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: Known-CVE Exploitation Specialist at [endpoint]
+- Severity: Critical
+- CWE: CWE-1395
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Depends on CVE — up to full compromise
+- Remediation: Patch/upgrade the affected components; apply vendor advisories
+```
+
+## System Prompt
+You are a specialist in exploiting known CVEs for the detected stack. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/drupal_audit.md

@@ -0,0 +1,36 @@
+# Drupal Security Audit Agent
+
+## User Prompt
+You are testing **{target}** for Drupal core/module weaknesses (e.g. Drupalgeddon class).
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Enumerate
+- Version (CHANGELOG, headers), enabled modules
+
+### 2. Correlate CVEs
+- Map to known Drupal RCE/SQLi (e.g. SA-CORE highly-critical classes)
+
+### 3. Confirm
+- Reproduce with an OOB/output proof where applicable
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: Drupal Security Audit at [endpoint]
+- Severity: Critical
+- CWE: CWE-1395
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Remote code execution
+- Remediation: Patch core/modules promptly
+```
+
+## System Prompt
+You are a specialist in Drupal core/module weaknesses (e.g. Drupalgeddon class). AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/git_svn_exposure_app.md

@@ -0,0 +1,36 @@
+# Exposed VCS / Build Artifacts Agent
+
+## User Prompt
+You are testing **{target}** for exposed .git/.svn/CI artifacts on the app host.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Probe
+- Request `/.git/HEAD`, `/.svn/entries`, `/.env`, build/CI artifact paths
+
+### 2. Recover
+- Dump source (git-dumper) / read secrets
+
+### 3. Confirm
+- Show recovered source or live secret
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: Exposed VCS / Build Artifacts at [endpoint]
+- Severity: High
+- CWE: CWE-527
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Source/secret disclosure → RCE
+- Remediation: Block VCS/dotfiles from web; rotate secrets
+```
+
+## System Prompt
+You are a specialist in exposed .git/.svn/CI artifacts on the app host. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/iis_handler_bypass.md

@@ -0,0 +1,36 @@
+# IIS Handler/Extension Bypass Agent
+
+## User Prompt
+You are testing **{target}** for auth or filter bypass via IIS handler quirks.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Probe
+- Test path/extension tricks: `;.asp`, `::$DATA`, trailing dot, `%20`, case, `/admin/.`/`..%2f`
+
+### 2. Bypass
+- Reach a protected handler/endpoint via a normalization or handler-mapping quirk
+
+### 3. Confirm
+- Show access to a resource that should be blocked
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: IIS Handler/Extension Bypass at [endpoint]
+- Severity: High
+- CWE: CWE-288
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Auth/control bypass
+- Remediation: Consistent normalization; patch; tighten ACLs
+```
+
+## System Prompt
+You are a specialist in auth or filter bypass via IIS handler quirks. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/iis_tilde_shortname.md

@@ -0,0 +1,37 @@
+# IIS Tilde (~) Short-Name Enumeration Agent
+
+## User Prompt
+You are testing **{target}** for IIS 8.3 short-name disclosure.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Detect
+- Probe `GET /*~1*/.aspx` style requests; a 404-vs-error differential reveals 8.3 short names
+- Confirm IIS version from Server header
+
+### 2. Enumerate
+- Brute the short names char by char to reveal hidden files/dirs
+
+### 3. Confirm
+- Show recovered short names mapping to real sensitive files
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: IIS Tilde (~) Short-Name Enumeration at [endpoint]
+- Severity: Medium
+- CWE: CWE-200
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Discovery of hidden files/backups/configs
+- Remediation: Disable 8.3 name creation; patch IIS
+```
+
+## System Prompt
+You are a specialist in IIS 8.3 short-name disclosure. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/iis_webdav.md

@@ -0,0 +1,36 @@
+# IIS WebDAV Misconfiguration Agent
+
+## User Prompt
+You are testing **{target}** for exposed/unsafe WebDAV on IIS.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Detect
+- `OPTIONS /` — look for DAV header / PUT/MOVE/COPY allowed
+
+### 2. Test write
+- Attempt PUT of a benign file; if blocked, try `.txt`→MOVE→`.asp` trick
+
+### 3. Confirm
+- Show an uploaded file is served (and if executable → RCE)
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: IIS WebDAV Misconfiguration at [endpoint]
+- Severity: High
+- CWE: CWE-650
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Arbitrary upload, potential RCE
+- Remediation: Disable WebDAV or restrict methods/authn
+```
+
+## System Prompt
+You are a specialist in exposed/unsafe WebDAV on IIS. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/joomla_audit.md

@@ -0,0 +1,36 @@
+# Joomla Security Audit Agent
+
+## User Prompt
+You are testing **{target}** for Joomla core/extension weaknesses.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Enumerate
+- Version (`administrator/manifests/files/joomla.xml`), components/extensions + versions
+
+### 2. Correlate CVEs
+- Map to known Joomla/extension CVEs (SQLi, LFI, object injection)
+
+### 3. Confirm
+- Reproduce one with proof
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: Joomla Security Audit at [endpoint]
+- Severity: High
+- CWE: CWE-1395
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Site takeover / data breach
+- Remediation: Update core/extensions; harden admin
+```
+
+## System Prompt
+You are a specialist in Joomla core/extension weaknesses. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/outdated_dependency_cve.md

@@ -0,0 +1,36 @@
+# Outdated Component CVE Specialist Agent
+
+## User Prompt
+You are testing **{target}** for outdated front-end/back-end components with known CVEs.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Inventory
+- Extract JS libs (jQuery, Angular, etc.), server modules, framework versions from responses/JS/headers
+
+### 2. Correlate
+- Map each to known CVEs; flag the exploitable, reachable ones
+
+### 3. Confirm
+- Prove exploitability where a safe PoC exists; else report as version-based exposure
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: Outdated Component CVE Specialist at [endpoint]
+- Severity: High
+- CWE: CWE-1104
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Varies — XSS/RCE/info-leak
+- Remediation: Upgrade components; dependency scanning in CI
+```
+
+## System Prompt
+You are a specialist in outdated front-end/back-end components with known CVEs. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

agents_md/vulns/wordpress_audit.md

@@ -0,0 +1,36 @@
+# WordPress Security Audit Agent
+
+## User Prompt
+You are testing **{target}** for WordPress core/plugin/theme weaknesses.
+
+**Recon Context:**
+{recon_json}
+
+**METHODOLOGY:**
+
+### 1. Enumerate
+- Users (`/?author=`, REST `/wp-json/wp/v2/users`), plugins/themes + versions, `xmlrpc.php`
+
+### 2. Correlate CVEs
+- Map plugin/theme versions to known vulns (arbitrary upload, SQLi, auth bypass, LFI)
+
+### 3. Confirm
+- Reproduce one concrete issue (e.g. unauth arbitrary file upload) with proof
+
+### 4. Report Format
+For each CONFIRMED finding:
+```
+FINDING:
+- Title: WordPress Security Audit at [endpoint]
+- Severity: High
+- CWE: CWE-1395
+- Endpoint: [full URL]
+- Vector: [what/where]
+- Payload: [exact payload/command]
+- Evidence: [raw tool output proving it]
+- Impact: Site takeover / RCE
+- Remediation: Update core/plugins/themes; harden; disable xmlrpc
+```
+
+## System Prompt
+You are a specialist in WordPress core/plugin/theme weaknesses. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. Confirm the component/version before claiming a version-specific CVE is exploitable; if you cannot reach a working PoC, report it as a lower-confidence exposure, not a confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.

neurosploit-rs/.neurosploit/history.txt

@@ -22,3 +22,10 @@ run
 /results
 /report
 /exit
+/run
+/status
+/help
+/results
+/stop
+/results
+/exit

neurosploit-rs/.neurosploit/runs.json

@@ -19,5 +19,12 @@
     "target": "http://testasp.vulnweb.com/",
     "workdir": "",
     "findings": []
+  },
+  {
+    "id": 4,
+    "mode": "black-box",
+    "target": "http://testasp.vulnweb.com/",
+    "workdir": "/opt/NeurosploitFinal/runs/ns-1782353097-testasp_vulnweb_com",
+    "findings": []
   }
 ]

neurosploit-rs/app/src/main.rs

@@ -359,6 +359,7 @@ pub(crate) struct Spawned {
     pub task: tokio::task::JoinHandle<RunOutput>,
     pub rx: tokio::sync::mpsc::Receiver<String>,
     pub cancel: std::sync::Arc<std::sync::atomic::AtomicBool>,
+    pub soft: std::sync::Arc<std::sync::atomic::AtomicBool>,
     pub workdir: PathBuf,
 }
 
@@ -408,6 +409,7 @@ pub(crate) fn spawn_engagement(base: &Path, mut cfg: RunConfig, mcp: bool, mode:
     let refs: Vec<ModelRef> = cfg.models.iter().map(|s| ModelRef::parse(s)).collect();
     let pool = ModelPool::with_auth(refs, cfg.concurrency, cfg.subscription, mcp_config);
     let cancel = pool.cancel_handle();
+    let soft = pool.soft_handle();
     let (tx, rx) = tokio::sync::mpsc::channel::<String>(256);
     let task = tokio::spawn(async move {
         match mode {
@@ -417,7 +419,25 @@ pub(crate) fn spawn_engagement(base: &Path, mut cfg: RunConfig, mcp: bool, mode:
             Mode::Black => harness::run(cfg, &lib, &pool, tx).await,
         }
     });
-    Spawned { task, rx, cancel, workdir }
+    Spawned { task, rx, cancel, soft, workdir }
+}
+
+/// Absolute file:// URL of a run's report (PDF if present, else HTML).
+pub(crate) fn report_url(workdir: &Path) -> String {
+    let pdf = workdir.join("report.pdf");
+    let f = if pdf.is_file() { pdf } else { workdir.join("report.html") };
+    let abs = f.canonicalize().unwrap_or(f);
+    format!("file://{}", abs.display())
+}
+
+/// Generate a report directly from raw (unvalidated) findings — used by the REPL
+/// when the user chooses "report without validating" on /stop.
+pub(crate) fn report_raw(target: &str, findings: &[harness::types::Finding], workdir: &Path) {
+    let mut fs = findings.to_vec();
+    harness::attack_graph::enrich(&mut fs);
+    std::fs::write(workdir.join("findings.json"), serde_json::to_string_pretty(&fs).unwrap_or_default()).ok();
+    let _ = harness::report::typst_report(target, &fs, workdir);
+    write_status(workdir, "stopped-raw", &format!("\"findings\":{}", fs.len()));
 }
 
 /// Generate the report + final status for a finished run, ensuring the workdir
@@ -433,7 +453,7 @@ pub(crate) fn finalize_run(mut out: RunOutput, workdir: &Path) -> RunOutput {
 }
 
 async fn run_mode(base: &Path, cfg: RunConfig, mcp: bool, mode: Mode) -> anyhow::Result<RunOutput> {
-    let Spawned { mut task, mut rx, cancel, workdir } = spawn_engagement(base, cfg, mcp, mode);
+    let Spawned { mut task, mut rx, cancel, workdir, .. } = spawn_engagement(base, cfg, mcp, mode);
     let printer = tokio::spawn(async move {
         while let Some(line) = rx.recv().await { render_line(&line); }
     });
@@ -465,7 +485,8 @@ async fn run_mode(base: &Path, cfg: RunConfig, mcp: bool, mode: Mode) -> anyhow:
     }
 
     let out = finalize_run(out, &workdir);
-    println!("  ✓ COMPLETE — {} validated finding(s) · status: {}/status.json", out.findings.len(), workdir.display());
+    println!("  ✓ COMPLETE — {} validated finding(s)", out.findings.len());
+    println!("  \x1b[36mreport: {}\x1b[0m", report_url(&workdir));
     Ok(out)
 }
 
@@ -560,6 +581,7 @@ pub(crate) fn render_compact(raw: &str) -> Option<String> {
         if let Some((label, rest)) = stripped.split_once(' ') { who = format!("[{label}] "); line = rest; }
     }
     let (tag, rest) = line.split_once(": ").unwrap_or(("", line));
+    if tag == "finding_json" { return None; } // captured for /results & /finding, not shown
     let s = match tag {
         "exec" | "danger" => format!("\x1b[33m  ⌘ {who}{}\x1b[0m", trunc1(rest, 110)),
         "net" => format!("\x1b[36m  🌐 {who}{}\x1b[0m", trunc1(rest, 110)),

neurosploit-rs/app/src/repl.rs

@@ -13,7 +13,7 @@ use rustyline::highlight::Highlighter;
 use rustyline::hint::Hinter;
 use rustyline::history::FileHistory;
 use rustyline::validate::{ValidationContext, ValidationResult, Validator};
-use rustyline::{CompletionType, Config, Context, Editor, ExternalPrinter, Helper};
+use rustyline::{Cmd, CompletionType, Config, Context, Editor, ExternalPrinter, Helper, KeyEvent};
 use serde::{Deserialize, Serialize};
 use std::io::IsTerminal;
 use std::path::Path;
@@ -28,7 +28,9 @@ struct RunLive {
     mode: &'static str,
     phase: String,
     started: Instant,
-    findings: Vec<(String, String)>, // sev, title
+    findings: Vec<(String, String)>, // sev, title (summary)
+    full: Vec<Finding>,              // full candidate findings (PoC, evidence) for /finding
+    commands: Vec<String>,           // full untruncated commands for /expand & Ctrl+O
     agents: usize,
     agents_done: usize,
 }
@@ -67,14 +69,30 @@ impl RunLive {
                 }
             }
         }
+        // Full candidate finding (with PoC/evidence) for /results & /finding.
+        if let Some(j) = line.strip_prefix("finding_json: ") {
+            if let Ok(f) = serde_json::from_str::<Finding>(j) { self.full.push(f); }
+        }
+        // Full untruncated command for /expand & Ctrl+O.
+        let cmd_part = line.strip_prefix('@').and_then(|s| s.split_once(' ').map(|(_, r)| r)).unwrap_or(line);
+        if let Some(c) = cmd_part.strip_prefix("exec: ").or_else(|| cmd_part.strip_prefix("danger: ")) {
+            self.commands.push(c.to_string());
+            if self.commands.len() > 100 { self.commands.remove(0); }
+        }
     }
 }
 
+/// What to do when the user stops a run.
+#[derive(Clone, Copy, PartialEq)]
+enum StopMode { Run, Validate, Raw, Discard }
+
 /// A run executing in the background of the REPL.
 struct ActiveRun {
     live: Arc<Mutex<RunLive>>,
     cancel: Arc<AtomicBool>,
+    soft: Arc<AtomicBool>,
     done: Arc<AtomicBool>,
+    choice: Arc<Mutex<StopMode>>,
 }
 
 /// All slash-commands, for Tab completion.
@@ -205,6 +223,8 @@ impl Reader {
                 .completion_type(CompletionType::List).build();
             if let Ok(mut ed) = Editor::<NsHelper, FileHistory>::with_config(cfg) {
                 ed.set_helper(Some(NsHelper));
+                // Ctrl+O pre-fills /expand to dump the last full (untruncated) commands.
+                ed.bind_sequence(KeyEvent::ctrl('o'), Cmd::Insert(1, "/expand".to_string()));
                 let hist = proj_dir().join("history.txt");
                 let _ = ed.load_history(&hist);
                 return Reader::Rl(Box::new(ed), hist);
@@ -376,7 +396,21 @@ pub async fn repl(base: &Path) -> anyhow::Result<()> {
             }
             "/stop" => {
                 match &active {
-                    Some(a) if !a.done.load(Ordering::Relaxed) => { a.cancel.store(true, Ordering::Relaxed); println!("  ⏸ stopping — finishing in-flight work; a report is generated on completion."); }
+                    Some(a) if !a.done.load(Ordering::Relaxed) => {
+                        println!("  \x1b[1mStop the run — choose:\x1b[0m");
+                        println!("    \x1b[36m1\x1b[0m  validate the findings found so far, then report  \x1b[2m(recommended)\x1b[0m");
+                        println!("    \x1b[36m2\x1b[0m  report NOW without validating (raw findings)");
+                        println!("    \x1b[36m3\x1b[0m  discard (no report)");
+                        let ans = ask_line("  choice [1/2/3]:");
+                        match ans.trim() {
+                            "2" => { *a.choice.lock().unwrap() = StopMode::Raw; a.cancel.store(true, Ordering::Relaxed);
+                                     println!("  ⏹ stopping — generating a RAW report from what was found…"); }
+                            "3" => { *a.choice.lock().unwrap() = StopMode::Discard; a.cancel.store(true, Ordering::Relaxed);
+                                     println!("  🗑 stopping — discarding this run."); }
+                            _   => { *a.choice.lock().unwrap() = StopMode::Validate; a.soft.store(true, Ordering::Relaxed);
+                                     println!("  ⏸ stopping exploitation — validating what was found, then reporting…"); }
+                        }
+                    }
                     _ => println!("  no active run."),
                 }
             }
@@ -394,7 +428,44 @@ pub async fn repl(base: &Path) -> anyhow::Result<()> {
                     println!("  ↻ retest set up for {} ({} prior finding(s)) — /run to launch", r.target, titles.len());
                 }
             }
-            "/results" => results(&history.lock().unwrap(), arg),
+            "/results" => {
+                // Live findings while a run is active (no arg), else a past run.
+                match &active {
+                    Some(a) if arg.is_empty() && !a.done.load(Ordering::Relaxed) => {
+                        let l = a.live.lock().unwrap();
+                        println!("  ▶ live — {} possible finding(s) so far ({})", l.full.len(), l.phase);
+                        let mut f = l.full.clone();
+                        f.sort_by_key(|x| sev_rank(&x.severity));
+                        for x in &f { println!("  • [{}] {} \x1b[2m({} · {})\x1b[0m", x.severity, x.title, x.agent, x.endpoint); }
+                        if !f.is_empty() { println!("  \x1b[2m/finding — pick one to see the command & PoC\x1b[0m"); }
+                    }
+                    _ => results(&history.lock().unwrap(), arg),
+                }
+            }
+            "/finding" | "/findings" => {
+                // Build the finding pool: live run if active, else a past run.
+                let pool: Vec<Finding> = match &active {
+                    Some(a) if arg.is_empty() && !a.done.load(Ordering::Relaxed) => a.live.lock().unwrap().full.clone(),
+                    _ => { let h = history.lock().unwrap(); pick(&h, arg).map(|r| r.findings.clone()).unwrap_or_default() }
+                };
+                finding_detail(&pool);
+            }
+            "/expand" | "/full" => {
+                // Show full untruncated commands from the active run.
+                match &active {
+                    Some(a) => {
+                        let l = a.live.lock().unwrap();
+                        let n: usize = arg.trim().parse().unwrap_or(5);
+                        let cmds = &l.commands;
+                        if cmds.is_empty() { println!("  no commands captured yet."); }
+                        else {
+                            println!("  ── last {} command(s) (full) ──", n.min(cmds.len()));
+                            for c in cmds.iter().rev().take(n).rev() { println!("  \x1b[33m$ {c}\x1b[0m"); }
+                        }
+                    }
+                    None => println!("  no active run — /expand shows full commands while a run streams."),
+                }
+            }
             "/report" => open_report(&history.lock().unwrap(), arg),
             "/status" => {
                 // Live status if a run is active, else a past run's status.json.
@@ -585,11 +656,14 @@ async fn start_background(base: &Path, s: &Session, reader: &mut Reader,
 
     let live = Arc::new(Mutex::new(RunLive {
         target: target.clone(), mode: mode_s, phase: "starting".into(),
-        started: Instant::now(), findings: vec![], agents: 0, agents_done: 0,
+        started: Instant::now(), findings: vec![], full: vec![], commands: vec![],
+        agents: 0, agents_done: 0,
     }));
     let cancel = sp.cancel.clone();
+    let soft = sp.soft.clone();
     let done = Arc::new(AtomicBool::new(false));
-    let (live2, done2, hist2) = (live.clone(), done.clone(), history);
+    let choice = Arc::new(Mutex::new(StopMode::Run));
+    let (live2, done2, hist2, choice2) = (live.clone(), done.clone(), history, choice.clone());
 
     tokio::spawn(async move {
         let crate::Spawned { task, mut rx, workdir, .. } = sp;
@@ -597,20 +671,40 @@ async fn start_background(base: &Path, s: &Session, reader: &mut Reader,
             live2.lock().unwrap().ingest(&line);
             if let Some(out) = crate::render_compact(&line) { let _ = printer.print(out); }
         }
-        let out = crate::finalize_run(task.await.unwrap_or_default(), &workdir);
+        let task_out = task.await.unwrap_or_default();
+        let mode_choice = *choice2.lock().unwrap();
+
+        if mode_choice == StopMode::Discard {
+            std::fs::remove_dir_all(&workdir).ok();
+            let _ = printer.print(format!("\x1b[33m🗑 run discarded — {}\x1b[0m", workdir.display()));
+            done2.store(true, Ordering::Relaxed);
+            return;
+        }
+
+        // Raw → report from the unvalidated candidates we captured live.
+        let (findings, validated_word) = if mode_choice == StopMode::Raw {
+            let raw = live2.lock().unwrap().full.clone();
+            crate::report_raw(&target, &raw, &workdir);
+            (raw, "unvalidated")
+        } else {
+            let out = crate::finalize_run(task_out, &workdir);
+            (out.findings, "validated")
+        };
+
         let id = {
             let mut h = hist2.lock().unwrap();
             let id = h.len() + 1;
-            h.push(RunRecord { id, mode: mode_s.into(), target, workdir: out.workdir.clone(), findings: out.findings.clone() });
+            h.push(RunRecord { id, mode: mode_s.into(), target, workdir: workdir.display().to_string(), findings: findings.clone() });
             if let Ok(j) = serde_json::to_string_pretty(&*h) { std::fs::write(proj_dir().join("runs.json"), j).ok(); }
             id
         };
         let _ = printer.print(format!(
-            "\x1b[1;32m◀ run #{id} done — {} validated finding(s)\x1b[0m · /results {id} · /report {id}",
+            "\x1b[1;32m◀ run #{id} done — {} {} finding(s)\x1b[0m · /results {id} · /finding",
-            out.findings.len()));
+            findings.len(), validated_word));
+        let _ = printer.print(format!("\x1b[36m  report: {}\x1b[0m", crate::report_url(&workdir)));
         done2.store(true, Ordering::Relaxed);
     });
-    Some(ActiveRun { live, cancel, done })
+    Some(ActiveRun { live, cancel, soft, done, choice })
 }
 
 /// Project-local store: `<cwd>/.neurosploit/` so each project keeps its own
@@ -737,6 +831,51 @@ fn diff_runs(history: &[RunRecord]) {
     if a == b { println!("  (no change in finding titles)"); }
 }
 
+fn sev_rank(s: &str) -> u8 {
+    match s { "Critical" => 0, "High" => 1, "Medium" => 2, "Low" => 3, _ => 4 }
+}
+
+/// Read one line synchronously (for the /stop choice prompt).
+fn ask_line(prompt: &str) -> String {
+    use std::io::Write;
+    print!("{prompt} ");
+    std::io::stdout().flush().ok();
+    let mut s = String::new();
+    std::io::stdin().read_line(&mut s).ok();
+    s
+}
+
+/// Arrow-key selection menu over findings; prints EVERYTHING about the chosen one
+/// (command/PoC, evidence, impact, remediation, votes, confidence).
+fn finding_detail(pool: &[Finding]) {
+    if pool.is_empty() { println!("  no findings to inspect yet."); return; }
+    let mut f = pool.to_vec();
+    f.sort_by_key(|x| sev_rank(&x.severity));
+    let items: Vec<String> = f.iter().map(|x| format!("[{}] {} — {}", x.severity, x.title, x.cwe)).collect();
+    let idx = if std::io::stdin().is_terminal() {
+        match dialoguer::Select::with_theme(&ColorfulTheme::default())
+            .with_prompt("Select a finding (↑/↓, enter)").items(&items).default(0).interact_opt() {
+            Ok(Some(i)) => i, _ => return,
+        }
+    } else { 0 };
+    let x = &f[idx];
+    println!("\n  ┌─ \x1b[1m{}\x1b[0m", x.title);
+    println!("  │  severity   : {}", x.severity);
+    println!("  │  cwe / cvss : {} · {}", x.cwe, x.cvss);
+    println!("  │  agent      : {}", x.agent);
+    println!("  │  endpoint   : {}", x.endpoint);
+    println!("  │  votes/conf : {} · {:.2}", x.votes, x.confidence);
+    println!("  ├─ \x1b[33mPayload / PoC\x1b[0m");
+    for l in x.payload.lines() { println!("  │  {l}"); }
+    println!("  ├─ \x1b[36mEvidence (tool output)\x1b[0m");
+    for l in x.evidence.lines() { println!("  │  {l}"); }
+    println!("  ├─ Impact");
+    for l in x.impact.lines() { println!("  │  {l}"); }
+    println!("  ├─ Remediation");
+    for l in x.remediation.lines() { println!("  │  {l}"); }
+    println!("  └─────");
+}
+
 fn run_status(history: &[RunRecord], arg: &str) {
     let Some(r) = pick(history, arg) else { return };
     match std::fs::read_to_string(Path::new(&r.workdir).join("status.json")) {

neurosploit-rs/crates/harness/src/pipeline.rs

@@ -159,7 +159,7 @@ pub async fn run(cfg: RunConfig, lib: &Library, pool: &ModelPool, tx: Sender<Str
             let directives = directives.clone();
             let txc = tx.clone();
             async move {
-                if pool.is_cancelled() {
+                if pool.stop_exploiting() {
                     return (ag.name.clone(), String::new(), vec![]);
                 }
                 if verbose {
@@ -184,6 +184,7 @@ pub async fn run(cfg: RunConfig, lib: &Library, pool: &ModelPool, tx: Sender<Str
                         // Live findings feed: surface each candidate the moment it appears.
                         for c in &f {
                             let _ = txc.send(format!("finding: [{}] {} @ {}", c.severity, c.title, c.endpoint)).await;
+                            if let Ok(j) = serde_json::to_string(c) { let _ = txc.send(format!("finding_json: {j}")).await; }
                         }
                         (ag.name.clone(), text, f)
                     }
@@ -378,7 +379,7 @@ pub async fn run_greybox(cfg: RunConfig, lib: &Library, pool: &ModelPool, tx: Se
             let leads = leads_ctx.clone();
             let txc = tx.clone();
             async move {
-                if pool.is_cancelled() {
+                if pool.stop_exploiting() {
                     return (ag.name.clone(), String::new(), vec![]);
                 }
                 if verbose {
@@ -931,7 +932,7 @@ pub async fn run_host(cfg: RunConfig, lib: &Library, pool: &ModelPool, tx: Sende
             let directives = directives.clone();
             let txc = tx.clone();
             async move {
-                if pool.is_cancelled() { return (ag.name.clone(), String::new(), vec![]); }
+                if pool.stop_exploiting() { return (ag.name.clone(), String::new(), vec![]); }
                 if verbose {
                     let _ = txc.send(format!("  ▶ launching agent: {} ({})", ag.name, ag.title.replace(" Agent", ""))).await;
                 }
@@ -944,7 +945,10 @@ pub async fn run_host(cfg: RunConfig, lib: &Library, pool: &ModelPool, tx: Sende
                     Ok((m, text)) => {
                         let f = extract_findings(&text, &ag.name);
                         let _ = txc.send(format!("test {} via {} → {} candidate(s)", ag.name, m.label(), f.len())).await;
-                        for c in &f { let _ = txc.send(format!("finding: [{}] {} @ {}", c.severity, c.title, c.endpoint)).await; }
+                        for c in &f {
+                            let _ = txc.send(format!("finding: [{}] {} @ {}", c.severity, c.title, c.endpoint)).await;
+                            if let Ok(j) = serde_json::to_string(c) { let _ = txc.send(format!("finding_json: {j}")).await; }
+                        }
                         (ag.name.clone(), text, f)
                     }
                     Err(e) => { let _ = txc.send(format!("test {} failed: {e}", ag.name)).await;

neurosploit-rs/crates/harness/src/pool.rs

@@ -34,9 +34,11 @@ pub struct ModelPool {
     /// Progress channel: when set, the subscription CLI streams structured
     /// activity (tools called, commands run, files read) here live.
     progress: std::sync::Mutex<Option<tokio::sync::mpsc::Sender<String>>>,
-    /// Cooperative cancellation: when set, in-flight model calls short-circuit
+    /// HARD cancellation: when set, in-flight model calls short-circuit (abort).
-    /// and the pipeline stops launching new agents (graceful stop).
     cancel: std::sync::Arc<std::sync::atomic::AtomicBool>,
+    /// SOFT stop: stop launching new EXPLOIT agents, but let in-flight finish and
+    /// VALIDATION still run — so "stop and validate what was found" works.
+    soft: std::sync::Arc<std::sync::atomic::AtomicBool>,
 }
 
 impl ModelPool {
@@ -65,6 +67,7 @@ impl ModelPool {
             mcp_config,
             progress: std::sync::Mutex::new(None),
             cancel: Arc::new(std::sync::atomic::AtomicBool::new(false)),
+            soft: Arc::new(std::sync::atomic::AtomicBool::new(false)),
         }
     }
 
@@ -80,13 +83,23 @@ impl ModelPool {
         self.progress.lock().ok().and_then(|g| g.clone())
     }
 
-    /// Handle to request graceful cancellation of an in-progress engagement.
+    /// Handle to request HARD cancellation (abort all model calls).
     pub fn cancel_handle(&self) -> Arc<std::sync::atomic::AtomicBool> {
         self.cancel.clone()
     }
+    /// Handle to request a SOFT stop (stop launching new exploit agents; keep
+    /// validation running).
+    pub fn soft_handle(&self) -> Arc<std::sync::atomic::AtomicBool> {
+        self.soft.clone()
+    }
     pub fn is_cancelled(&self) -> bool {
         self.cancel.load(std::sync::atomic::Ordering::Relaxed)
     }
+    /// Should the exploit phase stop launching new agents? (hard OR soft stop)
+    pub fn stop_exploiting(&self) -> bool {
+        self.cancel.load(std::sync::atomic::Ordering::Relaxed)
+            || self.soft.load(std::sync::atomic::Ordering::Relaxed)
+    }
 
     /// One completion for a model, via subscription CLI (optionally with MCP) or
     /// HTTP API, with a short retry/backoff. `label` (e.g. the agent name) tags

neurosploit-rs/crates/harness/src/report.rs

@@ -139,9 +139,10 @@ pub fn typst_report(target: &str, findings: &[Finding], dir: &Path) -> std::io::
     ));
     data.push_str("#let findings = (\n");
     for f in sorted_findings(findings) {
+        let owasp = if f.owasp.is_empty() { f.cwe.clone() } else { f.owasp.clone() };
         data.push_str(&format!(
-            "  (severity: {}, title: {}, agent: {}, cwe: {}, cvss: {}, endpoint: {}, payload: {}, evidence: {}, impact: {}, remediation: {}, votes: {}, confidence: {}),\n",
+            "  (severity: {}, title: {}, agent: {}, cwe: {}, owasp: {}, cvss: {}, endpoint: {}, payload: {}, evidence: {}, impact: {}, remediation: {}, votes: {}, confidence: {}),\n",
-            tq(&f.severity), tq(&f.title), tq(&f.agent), tq(&f.cwe), tq(&f.cvss),
+            tq(&f.severity), tq(&f.title), tq(&f.agent), tq(&f.cwe), tq(&owasp), tq(&f.cvss),
             tq(&f.endpoint), tq(&f.payload), tq(&f.evidence), tq(&f.impact),
             tq(&f.remediation), tq(&f.votes), f.confidence,
         ));

neurosploit-rs/templates/report.typ

@@ -71,13 +71,32 @@
   )
 ]
 
+#let sorted = findings.sorted(key: f => sevrank(f.severity))
+
+// ---- Vulnerability summary table ----
+#if sorted.len() > 0 [
+  #v(8pt)
+  == Vulnerability Summary
+  #v(4pt)
+  #table(
+    columns: (auto, 1fr, auto, auto, auto),
+    inset: 6pt, align: (left + horizon, left + horizon, center + horizon, center + horizon, center + horizon),
+    stroke: 0.5pt + rgb("#dddddd"),
+    table.header(
+      text(weight: "bold")[\#], text(weight: "bold")[Vulnerability],
+      text(weight: "bold")[Severity], text(weight: "bold")[CVSS], text(weight: "bold")[OWASP / CWE],
+    ),
+    ..sorted.enumerate().map(((i, f)) => (
+      str(i + 1), f.title, sevbadge(f.severity), f.cvss, f.owasp,
+    )).flatten()
+  )
+]
+
 #v(10pt)
 #line(length: 100%, stroke: 0.5pt + gray)
 
-// ---- Findings ----
+// ---- Detailed findings ----
 = Findings
-
-#let sorted = findings.sorted(key: f => sevrank(f.severity))
 #if sorted.len() == 0 [
   #text(fill: gray)[_Nothing to report._]
 ]
@@ -86,14 +105,20 @@
     stroke: (left: 3pt + sevcolor.at(f.severity, default: gray), rest: 0.5pt + rgb("#dddddd")))[
     #sevbadge(f.severity) #h(6pt) #text(12pt, weight: "bold")[#str(i + 1). #f.title]
     #v(4pt)
-    #text(9pt, fill: gray)[
+    #table(
-      agent: #raw(f.agent) · CWE: #f.cwe · CVSS: #f.cvss · votes: #f.votes · confidence: #str(f.confidence)
+      columns: (auto, 1fr, auto, 1fr),
-    ]
+      inset: 4pt, stroke: none, align: left + horizon,
-    #v(2pt) #text(9pt)[Endpoint: #raw(f.endpoint)]
+      text(8pt, fill: gray)[Criticality], text(8pt)[#f.severity],
-    #v(5pt) #strong[Payload] #linebreak() #raw(f.payload)
+      text(8pt, fill: gray)[CVSS], text(8pt)[#f.cvss],
+      text(8pt, fill: gray)[OWASP/CWE], text(8pt)[#f.owasp · #f.cwe],
+      text(8pt, fill: gray)[Confidence], text(8pt)[#f.votes votes · #str(f.confidence)],
+      text(8pt, fill: gray)[Location], text(8pt)[#raw(f.endpoint)],
+      text(8pt, fill: gray)[Agent], text(8pt)[#raw(f.agent)],
+    )
+    #v(4pt) #strong[Description / Impact] #linebreak() #text(9pt)[#f.impact]
+    #v(4pt) #strong[Proof of Concept] #linebreak() #raw(f.payload)
     #v(3pt) #strong[Evidence] #linebreak() #raw(f.evidence)
-    #v(3pt) #strong[Impact:] #f.impact
+    #v(3pt) #strong[Remediation] #linebreak() #text(9pt)[#f.remediation]
-    #v(2pt) #strong[Remediation:] #f.remediation
   ]
   #v(8pt)
 ]

scripts/build_appstack_agents.py

@@ -0,0 +1,143 @@
+#!/usr/bin/env python3
+"""
+NeuroSploit v3.5.1 — application-stack & CVE-hunting agents.
+Adds IIS/.NET, CMS (WordPress/Drupal/Joomla/etc.), app-server and known-CVE
+exploitation agents to agents_md/vulns/. Credits: Joas A Santos & Red Team Leaders.
+"""
+import os
+ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+OUT = os.path.join(ROOT, "agents_md", "vulns")
+
+
+def render(a):
+    L = [f"# {a['title']} Agent\n", "## User Prompt",
+         f"You are testing **{{target}}** for {a['for']}.\n",
+         "**Recon Context:**\n{recon_json}\n", "**METHODOLOGY:**\n"]
+    for i, (s, bs) in enumerate(a["steps"], 1):
+        L.append(f"### {i}. {s}")
+        L += [f"- {b}" for b in bs]
+        L.append("")
+    n = len(a["steps"]) + 1
+    L += [f"### {n}. Report Format", "For each CONFIRMED finding:", "```", "FINDING:",
+          f"- Title: {a['title']} at [endpoint]", f"- Severity: {a['sev']}", f"- CWE: {a['cwe']}",
+          "- Endpoint: [full URL]", "- Vector: [what/where]", "- Payload: [exact payload/command]",
+          "- Evidence: [raw tool output proving it]", f"- Impact: {a['impact']}",
+          f"- Remediation: {a['fix']}", "```\n", "## System Prompt", a["system"]]
+    return "\n".join(L) + "\n"
+
+
+def A(name, title, vc, cwe, sev, steps, fix, impact):
+    return {"name": name, "title": title, "for": vc, "sev": sev, "cwe": cwe, "impact": impact,
+            "fix": fix, "steps": steps,
+            "system": (f"You are a specialist in {vc}. AUTHORIZED engagement. Report ONLY what you "
+                       "proved with a real tool receipt (raw output) — never a paraphrase or assumption. "
+                       "Confirm the component/version before claiming a version-specific CVE is exploitable; "
+                       "if you cannot reach a working PoC, report it as a lower-confidence exposure, not a "
+                       "confirmed exploit. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.")}
+
+
+AGENTS = [
+ # ---- IIS / ASP.NET ----
+ A("iis_tilde_shortname", "IIS Tilde (~) Short-Name Enumeration", "IIS 8.3 short-name disclosure",
+   "CWE-200", "Medium",
+   [("Detect", ["Probe `GET /*~1*/.aspx` style requests; a 404-vs-error differential reveals 8.3 short names",
+                "Confirm IIS version from Server header"]),
+    ("Enumerate", ["Brute the short names char by char to reveal hidden files/dirs"]),
+    ("Confirm", ["Show recovered short names mapping to real sensitive files"])],
+   "Disable 8.3 name creation; patch IIS", "Discovery of hidden files/backups/configs"),
+ A("iis_webdav", "IIS WebDAV Misconfiguration", "exposed/unsafe WebDAV on IIS",
+   "CWE-650", "High",
+   [("Detect", ["`OPTIONS /` — look for DAV header / PUT/MOVE/COPY allowed"]),
+    ("Test write", ["Attempt PUT of a benign file; if blocked, try `.txt`→MOVE→`.asp` trick"]),
+    ("Confirm", ["Show an uploaded file is served (and if executable → RCE)"])],
+   "Disable WebDAV or restrict methods/authn", "Arbitrary upload, potential RCE"),
+ A("aspnet_viewstate", "ASP.NET ViewState Deserialization", "unprotected/known-key __VIEWSTATE deserialization",
+   "CWE-502", "Critical",
+   [("Inspect", ["Capture __VIEWSTATE; check if MAC is disabled (enableViewStateMac=false) or a known/leaked machineKey is in play"]),
+    ("Weaponize", ["With a known/guessed machineKey, craft a ysoserial.net ViewState gadget"]),
+    ("Confirm", ["Prove code execution via OOB callback or command output tied to a unique marker"])],
+   "Enable ViewState MAC; rotate machineKey; patch", "Remote code execution"),
+ A("aspnet_debug_trace", "ASP.NET Debug/Trace Exposure", "debug/trace enabled in production ASP.NET",
+   "CWE-489", "Medium",
+   [("Probe", ["Request `trace.axd`; send `DEBUG` verb; check `<compilation debug=...>` leakage via errors"]),
+    ("Assess", ["Harvest request/session data, stack traces, app internals from trace output"]),
+    ("Confirm", ["Show sensitive runtime data exposed"])],
+   "Disable debug/trace; custom errors", "Information disclosure"),
+ A("iis_handler_bypass", "IIS Handler/Extension Bypass", "auth or filter bypass via IIS handler quirks",
+   "CWE-288", "High",
+   [("Probe", ["Test path/extension tricks: `;.asp`, `::$DATA`, trailing dot, `%20`, case, `/admin/.`/`..%2f`"]),
+    ("Bypass", ["Reach a protected handler/endpoint via a normalization or handler-mapping quirk"]),
+    ("Confirm", ["Show access to a resource that should be blocked"])],
+   "Consistent normalization; patch; tighten ACLs", "Auth/control bypass"),
+ # ---- CMS general & specific ----
+ A("cms_fingerprint", "CMS Fingerprint & Version", "CMS identification and version disclosure",
+   "CWE-200", "Info",
+   [("Identify", ["Detect CMS via meta generator, paths (`/wp-`, `/sites/`, `/administrator/`), headers, favicon hash",
+                  "Run whatweb/wpscan-style detection without auth"]),
+    ("Version", ["Pin exact version from readme/changelog/asset hashes"]),
+    ("Map", ["List plugins/themes/modules and their versions for CVE correlation"])],
+   "Hide version/generator; keep components updated", "Targeted exploitation surface"),
+ A("wordpress_audit", "WordPress Security Audit", "WordPress core/plugin/theme weaknesses",
+   "CWE-1395", "High",
+   [("Enumerate", ["Users (`/?author=`, REST `/wp-json/wp/v2/users`), plugins/themes + versions, `xmlrpc.php`"]),
+    ("Correlate CVEs", ["Map plugin/theme versions to known vulns (arbitrary upload, SQLi, auth bypass, LFI)"]),
+    ("Confirm", ["Reproduce one concrete issue (e.g. unauth arbitrary file upload) with proof"])],
+   "Update core/plugins/themes; harden; disable xmlrpc", "Site takeover / RCE"),
+ A("joomla_audit", "Joomla Security Audit", "Joomla core/extension weaknesses",
+   "CWE-1395", "High",
+   [("Enumerate", ["Version (`administrator/manifests/files/joomla.xml`), components/extensions + versions"]),
+    ("Correlate CVEs", ["Map to known Joomla/extension CVEs (SQLi, LFI, object injection)"]),
+    ("Confirm", ["Reproduce one with proof"])],
+   "Update core/extensions; harden admin", "Site takeover / data breach"),
+ A("drupal_audit", "Drupal Security Audit", "Drupal core/module weaknesses (e.g. Drupalgeddon class)",
+   "CWE-1395", "Critical",
+   [("Enumerate", ["Version (CHANGELOG, headers), enabled modules"]),
+    ("Correlate CVEs", ["Map to known Drupal RCE/SQLi (e.g. SA-CORE highly-critical classes)"]),
+    ("Confirm", ["Reproduce with an OOB/output proof where applicable"])],
+   "Patch core/modules promptly", "Remote code execution"),
+ A("cms_default_admin", "CMS Admin Panel & Default Creds", "exposed CMS admin with weak/default credentials",
+   "CWE-1392", "High",
+   [("Locate", ["Find admin (`/wp-admin`, `/administrator`, `/user/login`, `/admin`)"]),
+    ("Test (in scope)", ["Try supplied/default credentials; respect lockout/ROE — no out-of-scope brute force"]),
+    ("Confirm", ["Show authenticated admin access"])],
+   "Remove defaults; strong creds + MFA; restrict admin", "Full CMS compromise"),
+ # ---- app servers / panels ----
+ A("appserver_exposure", "App-Server Console Exposure", "exposed Tomcat/JBoss/Jenkins/Actuator consoles",
+   "CWE-1188", "High",
+   [("Discover", ["Probe `/manager/html`, `/jmx-console`, `/jenkins`, `/actuator`, `/console`, `/admin`"]),
+    ("Assess", ["Test default/weak creds (in scope); check unauth-exposed management endpoints"]),
+    ("Confirm", ["Demonstrate a management action / deploy / info-leak proving exposure (→ often RCE)"])],
+   "Authenticate & network-restrict consoles; remove defaults", "Remote code execution / takeover"),
+ A("git_svn_exposure_app", "Exposed VCS / Build Artifacts", "exposed .git/.svn/CI artifacts on the app host",
+   "CWE-527", "High",
+   [("Probe", ["Request `/.git/HEAD`, `/.svn/entries`, `/.env`, build/CI artifact paths"]),
+    ("Recover", ["Dump source (git-dumper) / read secrets"]),
+    ("Confirm", ["Show recovered source or live secret"])],
+   "Block VCS/dotfiles from web; rotate secrets", "Source/secret disclosure → RCE"),
+ # ---- CVE hunting ----
+ A("cve_known_exploitation", "Known-CVE Exploitation Specialist", "exploiting known CVEs for the detected stack",
+   "CWE-1395", "Critical",
+   [("Identify versions", ["From recon, list each component + exact version (server, framework, CMS, plugins, libs)"]),
+    ("Map to CVEs", ["Match versions to known CVEs; prioritise unauth RCE/SQLi/auth-bypass; note CVE id + CVSS",
+                     "Prefer issues with a reliable, non-destructive PoC"]),
+    ("Reproduce safely", ["Run a benign PoC (e.g. a version/echo check or OOB callback) to confirm the CVE is actually present and exploitable — never a destructive payload"]),
+    ("Confirm", ["Report the CVE only when the PoC produced concrete proof (output/OOB); otherwise report it as 'potentially vulnerable (version match, unconfirmed)'"])],
+   "Patch/upgrade the affected components; apply vendor advisories", "Depends on CVE — up to full compromise"),
+ A("outdated_dependency_cve", "Outdated Component CVE Specialist", "outdated front-end/back-end components with known CVEs",
+   "CWE-1104", "High",
+   [("Inventory", ["Extract JS libs (jQuery, Angular, etc.), server modules, framework versions from responses/JS/headers"]),
+    ("Correlate", ["Map each to known CVEs; flag the exploitable, reachable ones"]),
+    ("Confirm", ["Prove exploitability where a safe PoC exists; else report as version-based exposure"])],
+   "Upgrade components; dependency scanning in CI", "Varies — XSS/RCE/info-leak"),
+]
+
+
+def main():
+    os.makedirs(OUT, exist_ok=True)
+    for a in AGENTS:
+        open(os.path.join(OUT, a["name"] + ".md"), "w").write(render(a))
+    print(f"wrote {len(AGENTS)} app-stack/CVE agents to {OUT}")
+
+
+if __name__ == "__main__":
+    main()