feat(3.6.6): local/uncensored llama.cpp provider, clippy clean, CI (#40)

Version bump 3.6.5 -> 3.6.6.

Local & uncensored models
- New `llamacpp:` provider (llama-server, OpenAI-compatible, localhost:8080,
  no API key, CPU-only or GPU-offloaded). Override via LLAMACPP_BASE_URL;
  model name is the loaded gguf (pass-through). 15 -> 16 providers.
- README: local/uncensored highlight, provider table + key-less note, badges.

Quality
- clippy clean under `-D warnings`: clamp(), sort_by_key(Reverse), struct-literal
  init, too_many_arguments allows, scoped await_holding_lock on the REPL blocking
  fallback (guard intentionally held across run().await), plus clippy --fix set.

CI
- examples/github-actions/ci.yml: cargo build/test/clippy -D warnings for the
  neurosploit-rs workspace (template, kept out of .github/workflows).


Claude-Session: https://claude.ai/code/session_01QDses7zTSa9YF7pPRjphvh

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Joas A Santos
2026-08-03 23:48:08 -03:00
committed by GitHub
co-authored by Claude Fable 5
parent 3786d7c559
commit f913af211d
14 changed files with 137 additions and 63 deletions
+16 -4
View File
@@ -1,4 +1,4 @@
<h1 align="center">🧠 NeuroSploit v3.6.5</h1> <h1 align="center">🧠 NeuroSploit v3.6.6</h1>
<p align="center"> <p align="center">
<a href="https://trendshift.io/repositories/22624?utm_source=trendshift-badge&amp;utm_medium=badge&amp;utm_campaign=badge-trendshift-22624" target="_blank" rel="noopener noreferrer"><img src="https://trendshift.io/api/badge/trendshift/repositories/22624/daily?language=Python" alt="JoasASantos%2FNeuroSploit | Trendshift" width="250" height="55"/></a> <a href="https://trendshift.io/repositories/22624?utm_source=trendshift-badge&amp;utm_medium=badge&amp;utm_campaign=badge-trendshift-22624" target="_blank" rel="noopener noreferrer"><img src="https://trendshift.io/api/badge/trendshift/repositories/22624/daily?language=Python" alt="JoasASantos%2FNeuroSploit | Trendshift" width="250" height="55"/></a>
@@ -12,11 +12,11 @@
</p> </p>
<p align="center"> <p align="center">
<img src="https://img.shields.io/badge/Version-3.6.5-blue?style=flat-square"> <img src="https://img.shields.io/badge/Version-3.6.6-blue?style=flat-square">
<img src="https://img.shields.io/badge/Harness-Rust%20%7C%20tokio-e6b673?style=flat-square"> <img src="https://img.shields.io/badge/Harness-Rust%20%7C%20tokio-e6b673?style=flat-square">
<img src="https://img.shields.io/badge/License-MIT-green?style=flat-square"> <img src="https://img.shields.io/badge/License-MIT-green?style=flat-square">
<img src="https://img.shields.io/badge/MD%20Agents-430-red?style=flat-square"> <img src="https://img.shields.io/badge/MD%20Agents-430-red?style=flat-square">
<img src="https://img.shields.io/badge/Models-15%20providers-success?style=flat-square"> <img src="https://img.shields.io/badge/Models-16%20providers-success?style=flat-square">
<img src="https://img.shields.io/badge/Modes-Black%20%7C%20White%20%7C%20Grey%20%7C%20Host%20%7C%20AI-9cf?style=flat-square"> <img src="https://img.shields.io/badge/Modes-Black%20%7C%20White%20%7C%20Grey%20%7C%20Host%20%7C%20AI-9cf?style=flat-square">
<img src="https://img.shields.io/badge/Auth-API%20key%20%7C%20Subscription-orange?style=flat-square"> <img src="https://img.shields.io/badge/Auth-API%20key%20%7C%20Subscription-orange?style=flat-square">
</p> </p>
@@ -108,6 +108,11 @@ Control TUI**.
- 📸 **Proof screenshots in reports** — agents capture visual proof per finding - 📸 **Proof screenshots in reports** — agents capture visual proof per finding
(`evidence/<finding-id>-N.png`), embedded beside its vulnerability in the (`evidence/<finding-id>-N.png`), embedded beside its vulnerability in the
Typst/HTML/Markdown reports. Typst/HTML/Markdown reports.
- 🖥️ **Local, uncensored & CPU-only models**`ollama:` and `llamacpp:` run the
whole engagement on your box with **no API key** and **no data leaving the
host**. `llamacpp:` speaks to a `llama-server` OpenAI-compatible endpoint
(`LLAMACPP_BASE_URL`, default localhost:8080); the `model` is whatever gguf you
loaded. Ideal for offline/air-gapped work and unfiltered offensive prompting.
- 🕵️ **Burp/ZAP proxy**`/proxy <url>` (or `/burp`) routes agent traffic - 🕵️ **Burp/ZAP proxy**`/proxy <url>` (or `/burp`) routes agent traffic
through your local intercepting proxy so you can inspect & replay in Burp. through your local intercepting proxy so you can inspect & replay in Burp.
- 🗺️ **Attack graph & kill chain** — findings mapped to OWASP / CWE / MITRE - 🗺️ **Attack graph & kill chain** — findings mapped to OWASP / CWE / MITRE
@@ -412,7 +417,7 @@ export GROQ_API_KEY=... # groq:*
export TOGETHER_API_KEY=... # together:* export TOGETHER_API_KEY=... # together:*
export MOONSHOT_API_KEY=... # moonshot:* (Kimi K3/K2) export MOONSHOT_API_KEY=... # moonshot:* (Kimi K3/K2)
export OPENROUTER_API_KEY=... # openrouter:* export OPENROUTER_API_KEY=... # openrouter:*
# ollama needs no key (local) # ollama / llamacpp need no key (local)
# then run via API (note: NO --subscription) # then run via API (note: NO --subscription)
./target/release/neurosploit run http://testphp.vulnweb.com/ \ ./target/release/neurosploit run http://testphp.vulnweb.com/ \
@@ -442,9 +447,16 @@ Or put the keys in a `.env` and source it (`cp .env.example .env`; edit; `set -a
| `moonshot:` | `MOONSHOT_API_KEY` | api.moonshot.ai | | `moonshot:` | `MOONSHOT_API_KEY` | api.moonshot.ai |
| `openrouter:` | `OPENROUTER_API_KEY` | openrouter.ai | | `openrouter:` | `OPENROUTER_API_KEY` | openrouter.ai |
| `ollama:` | _(none)_ | localhost:11434 | | `ollama:` | _(none)_ | localhost:11434 |
| `llamacpp:` | _(none)_ | localhost:8080 |
Run `./target/release/neurosploit models` for the full provider/model list. Run `./target/release/neurosploit models` for the full provider/model list.
> **Local, uncensored & CPU-only** — `ollama:` and `llamacpp:` run entirely on
> your box with no API key and no data leaving the host. `llamacpp:` targets a
> [`llama-server`](https://github.com/ggml-org/llama.cpp) OpenAI-compatible
> endpoint (override with `LLAMACPP_BASE_URL`); the `model` is whatever gguf you
> loaded. Ideal for offline engagements and unfiltered offensive prompting.
#### 2) Via subscription (no API key) #### 2) Via subscription (no API key)
`--subscription` drives your local agentic-CLI login instead of an API key — `--subscription` drives your local agentic-CLI login instead of an API key —
+1
View File
@@ -9,6 +9,7 @@ variables → Actions), or swap the `MODEL`/key for a provider you use. The buil
|----------|--------------| |----------|--------------|
| `neurosploit-pr-gate.yml` | Reviews every pull request and **blocks the merge** on a confirmed critical (fails the check + sets a `neurosploit/security` commit status + posts a REQUEST_CHANGES review). | | `neurosploit-pr-gate.yml` | Reviews every pull request and **blocks the merge** on a confirmed critical (fails the check + sets a `neurosploit/security` commit status + posts a REQUEST_CHANGES review). |
| `neurosploit-mention.yml` | Comment **`@neurosploit`** on a PR/issue (writers only) to trigger a scan. Text after the mention steers it, in any language; a URL runs a black-box test, otherwise it reviews the PR. | | `neurosploit-mention.yml` | Comment **`@neurosploit`** on a PR/issue (writers only) to trigger a scan. Text after the mention steers it, in any language; a URL runs a black-box test, otherwise it reviews the PR. |
| `ci.yml` | Rust CI for the `neurosploit-rs/` workspace — `cargo build` / `test` / `clippy -D warnings` on every push & PR. |
## Enforce the PR gate as a merge block ## Enforce the PR gate as a merge block
+52
View File
@@ -0,0 +1,52 @@
# NeuroSploit — Rust CI
#
# Build, test and lint the `neurosploit-rs/` Cargo workspace on every push and
# pull request. Copy to `.github/workflows/ci.yml` in your fork to enable it.
#
# Note: this lives here (not in `.github/workflows/`) as a template — like the
# other files in this folder — so the upstream repo doesn't run it on itself.
name: ci
on:
push:
branches: [main]
paths: ["neurosploit-rs/**", ".github/workflows/ci.yml"]
pull_request:
paths: ["neurosploit-rs/**"]
env:
CARGO_TERM_COLOR: always
RUST_BACKTRACE: 1
jobs:
build-test:
name: build & test
runs-on: ubuntu-latest
defaults:
run:
working-directory: neurosploit-rs
steps:
- uses: actions/checkout@v4
- name: Install Rust toolchain
run: rustup toolchain install stable --profile minimal --component clippy
- name: Cache cargo
uses: actions/cache@v4
with:
path: |
~/.cargo/registry
~/.cargo/git
neurosploit-rs/target
key: ${{ runner.os }}-cargo-${{ hashFiles('neurosploit-rs/Cargo.lock') }}
restore-keys: ${{ runner.os }}-cargo-
- name: Build
run: cargo build --workspace --locked
- name: Test
run: cargo test --workspace --locked
- name: Clippy
run: cargo clippy --workspace --all-targets -- -D warnings
+2 -2
View File
@@ -871,7 +871,7 @@ dependencies = [
[[package]] [[package]]
name = "neurosploit" name = "neurosploit"
version = "3.6.5" version = "3.6.6"
dependencies = [ dependencies = [
"anyhow", "anyhow",
"clap", "clap",
@@ -888,7 +888,7 @@ dependencies = [
[[package]] [[package]]
name = "neurosploit-harness" name = "neurosploit-harness"
version = "3.6.5" version = "3.6.6"
dependencies = [ dependencies = [
"anyhow", "anyhow",
"futures", "futures",
+1 -1
View File
@@ -3,7 +3,7 @@ members = ["crates/harness", "app"]
resolver = "2" resolver = "2"
[workspace.package] [workspace.package]
version = "3.6.5" version = "3.6.6"
edition = "2021" edition = "2021"
license = "MIT" license = "MIT"
repository = "https://github.com/JoasASantos/NeuroSploit" repository = "https://github.com/JoasASantos/NeuroSploit"
+5 -5
View File
@@ -1,4 +1,4 @@
//! NeuroSploit v3.6.5 — interactive harness + CLI (`run` / `whitebox` / `agents` / `models`). //! NeuroSploit v3.6.6 — interactive harness + CLI (`run` / `whitebox` / `agents` / `models`).
mod repl; mod repl;
mod tui; mod tui;
@@ -11,8 +11,8 @@ use std::path::{Path, PathBuf};
#[command( #[command(
name = "neurosploit", name = "neurosploit",
version, version,
about = "NeuroSploit v3.6.5 — multi-model autonomous pentest harness", about = "NeuroSploit v3.6.6 — multi-model autonomous pentest harness",
long_about = "NeuroSploit v3.6.5 — a Rust multi-model harness that drives a pool of LLMs \ long_about = "NeuroSploit v3.6.6 — a Rust multi-model harness that drives a pool of LLMs \
(API key or local subscription: Claude/Codex/Gemini/Grok) to autonomously test a target. \ (API key or local subscription: Claude/Codex/Gemini/Grok) to autonomously test a target. \
After recon it INTELLIGENTLY selects only the agents matching the discovered surface, runs \ After recon it INTELLIGENTLY selects only the agents matching the discovered surface, runs \
them in parallel, then validates every finding by cross-model voting before reporting.\n\n\ them in parallel, then validates every finding by cross-model voting before reporting.\n\n\
@@ -542,7 +542,7 @@ async fn main() -> anyhow::Result<()> {
std::fs::remove_dir_all(&dest).ok(); std::fs::remove_dir_all(&dest).ok();
let url = ig.authed_clone_url(&format!("https://github.com/{owner_repo}")); let url = ig.authed_clone_url(&format!("https://github.com/{owner_repo}"));
if run_git(&["clone", "--depth", "1", "--branch", &branch, &url, &dest.display().to_string()]).is_ok() { if run_git(&["clone", "--depth", "1", "--branch", &branch, &url, &dest.display().to_string()]).is_ok() {
let mut cfg = RunConfig::new(&dest.display().to_string()); let mut cfg = RunConfig::new(dest.display().to_string());
cfg.subscription = subscription; cfg.subscription = subscription;
cfg.verbose = verbose; cfg.verbose = verbose;
if !models.is_empty() { cfg.models = models.clone(); } if !models.is_empty() { cfg.models = models.clone(); }
@@ -751,7 +751,7 @@ pub(crate) fn spawn_engagement(base: &Path, mut cfg: RunConfig, mcp: bool, mode:
println!(" │ ua : {ua}"); println!(" │ ua : {ua}");
write_status(&workdir, "running", &format!("\"target\":{:?}", cfg.target)); write_status(&workdir, "running", &format!("\"target\":{:?}", cfg.target));
println!(" ┌─ NeuroSploit v3.6.5 · by Joas A Santos & Red Team Leaders"); println!(" ┌─ NeuroSploit v3.6.6 · by Joas A Santos & Red Team Leaders");
println!(" │ run id : {run_id}"); println!(" │ run id : {run_id}");
println!(" │ target : {}", cfg.target); println!(" │ target : {}", cfg.target);
println!(" │ models : {}", cfg.models.join(", ")); println!(" │ models : {}", cfg.models.join(", "));
+9 -8
View File
@@ -1,4 +1,4 @@
//! NeuroSploit v3.6.5 — interactive session (Claude-Code / Codex / Cursor-CLI style). //! NeuroSploit v3.6.6 — interactive session (Claude-Code / Codex / Cursor-CLI style).
//! //!
//! Launched when `neurosploit` runs with no subcommand. A persistent REPL with //! Launched when `neurosploit` runs with no subcommand. A persistent REPL with
//! real line editing (arrow-key history recall, Ctrl-A/E/K, paste), model //! real line editing (arrow-key history recall, Ctrl-A/E/K, paste), model
@@ -361,12 +361,16 @@ impl Reader {
} }
} }
// The blocking (piped, no external printer) fallback holds the history
// MutexGuard across `run().await` on purpose — run() mutates that history for
// the whole async operation and no other task contends for it there.
#[allow(clippy::await_holding_lock)]
pub async fn repl(base: &Path) -> anyhow::Result<()> { pub async fn repl(base: &Path) -> anyhow::Result<()> {
let lib = agents::load(base); let lib = agents::load(base);
let backends = harness::installed_cli_backends(); let backends = harness::installed_cli_backends();
println!("\x1b[1m"); println!("\x1b[1m");
println!(" ███╗ ██╗███████╗██╗ ██╗██████╗ ██████╗"); println!(" ███╗ ██╗███████╗██╗ ██╗██████╗ ██████╗");
println!(" ████╗ ██║██╔════╝██║ ██║██╔══██╗██╔═══██╗ NeuroSploit v3.6.5"); println!(" ████╗ ██║██╔════╝██║ ██║██╔══██╗██╔═══██╗ NeuroSploit v3.6.6");
println!(" ██╔██╗ ██║█████╗ ██║ ██║██████╔╝██║ ██║ interactive harness"); println!(" ██╔██╗ ██║█████╗ ██║ ██║██████╔╝██║ ██║ interactive harness");
println!(" ██║╚██╗██║██╔══╝ ██║ ██║██╔══██╗██║ ██║ by Joas A Santos"); println!(" ██║╚██╗██║██╔══╝ ██║ ██║██╔══██╗██║ ██║ by Joas A Santos");
println!(" ██║ ╚████║███████╗╚██████╔╝██║ ██║╚██████╔╝ & Red Team Leaders"); println!(" ██║ ╚████║███████╗╚██████╔╝██║ ██║╚██████╔╝ & Red Team Leaders");
@@ -1626,12 +1630,9 @@ fn browse_results(history: &[RunRecord]) {
}; };
print_finding_detail(&f[fi]); print_finding_detail(&f[fi]);
// Enter → back to the vuln list; Esc → back to the target list. // Enter → back to the vuln list; Esc → back to the target list.
match dialoguer::Select::with_theme(&ColorfulTheme::default()) if let Ok(None) = dialoguer::Select::with_theme(&ColorfulTheme::default())
.with_prompt("↵ back to vulnerabilities · Esc = back to targets") .with_prompt("↵ back to vulnerabilities · Esc = back to targets")
.items(&["back"]).default(0).interact_opt() { .items(&["back"]).default(0).interact_opt() { break }
Ok(None) => break,
_ => {}
}
} }
} }
} }
@@ -2118,7 +2119,7 @@ fn attach_path(spec: &str, s: &mut Session) -> usize {
Ok(content) => { Ok(content) => {
let body = match range.and_then(parse_range) { let body = match range.and_then(parse_range) {
Some((a, b)) => content.lines().enumerate() Some((a, b)) => content.lines().enumerate()
.filter(|(i, _)| *i + 1 >= a && *i + 1 <= b) .filter(|(i, _)| *i + 1 >= a && *i < b)
.map(|(_, l)| l).collect::<Vec<_>>().join("\n"), .map(|(_, l)| l).collect::<Vec<_>>().join("\n"),
None => content.chars().take(8000).collect(), None => content.chars().take(8000).collect(),
}; };
+4 -7
View File
@@ -1,4 +1,4 @@
//! NeuroSploit v3.6.5 — TUI "Mission Control" mode. //! NeuroSploit v3.6.6 — TUI "Mission Control" mode.
//! //!
//! Concurrent panels that update live while the engagement runs in the //! Concurrent panels that update live while the engagement runs in the
//! background, with a composer input that stays active during execution: //! background, with a composer input that stays active during execution:
@@ -169,7 +169,7 @@ pub async fn run(base: &Path, mut cfg: RunConfig, mcp: bool, mode: Mode) -> anyh
} }
}); });
let out;
loop { loop {
// drain engagement events // drain engagement events
while let Ok(line) = rx.try_recv() { ui.ingest(line); } while let Ok(line) = rx.try_recv() { ui.ingest(line); }
@@ -206,17 +206,14 @@ pub async fn run(base: &Path, mut cfg: RunConfig, mcp: bool, mode: Mode) -> anyh
} }
} }
out = (&mut task).await.unwrap_or_default(); let out = (&mut task).await.unwrap_or_default();
// ---- restore terminal ---- // ---- restore terminal ----
execute!(stdout(), terminal::LeaveAlternateScreen)?; execute!(stdout(), terminal::LeaveAlternateScreen)?;
terminal::disable_raw_mode()?; terminal::disable_raw_mode()?;
// generate report unless discarded; print a plain summary after leaving the TUI // generate report unless discarded; print a plain summary after leaving the TUI
match harness::report::typst_report(&out.target, &out.findings, &workdir) { if let Ok(p) = harness::report::typst_report(&out.target, &out.findings, &workdir) { println!(" report → {}", p.display()) }
Ok(p) => println!(" report → {}", p.display()),
Err(_) => {}
}
crate::write_status_pub(&workdir, if cancel.load(Ordering::Relaxed) { "stopped" } else { "complete" }, ""); crate::write_status_pub(&workdir, if cancel.load(Ordering::Relaxed) { "stopped" } else { "complete" }, "");
println!("{} validated finding(s) · {}", out.findings.len(), workdir.display()); println!("{} validated finding(s) · {}", out.findings.len(), workdir.display());
Ok(()) Ok(())
+5 -4
View File
@@ -147,10 +147,11 @@ pub fn hygiene_summary(findings: &[Finding]) -> Vec<String> {
mod tests { mod tests {
use super::*; use super::*;
fn f(title: &str, sev: &str, cwe: &str, ep: &str, ev: &str, payload: &str) -> Finding { fn f(title: &str, sev: &str, cwe: &str, ep: &str, ev: &str, payload: &str) -> Finding {
let mut x = Finding::default(); Finding {
x.title = title.into(); x.severity = sev.into(); x.cwe = cwe.into(); title: title.into(), severity: sev.into(), cwe: cwe.into(),
x.endpoint = ep.into(); x.evidence = ev.into(); x.payload = payload.into(); endpoint: ep.into(), evidence: ev.into(), payload: payload.into(),
x ..Default::default()
}
} }
#[test] #[test]
+9 -1
View File
@@ -60,6 +60,12 @@ pub fn providers() -> Vec<Provider> {
models: vec!["gpt-4o", "gpt-4o-mini", "gpt-5.1", "o4-mini"] }, models: vec!["gpt-4o", "gpt-4o-mini", "gpt-5.1", "o4-mini"] },
Provider { key: "ollama", label: "Ollama (local)", base_url: "http://localhost:11434/v1", env_key: "OLLAMA_API_KEY", kind: "api", Provider { key: "ollama", label: "Ollama (local)", base_url: "http://localhost:11434/v1", env_key: "OLLAMA_API_KEY", kind: "api",
models: vec!["qwen2.5-coder:32b", "qwq:32b", "deepseek-r1:32b", "llama3.3:70b"] }, models: vec!["qwen2.5-coder:32b", "qwq:32b", "deepseek-r1:32b", "llama3.3:70b"] },
// llama.cpp server (`llama-server`, OpenAI-compatible). Runs CPU-only or
// GPU-offloaded, fully local & uncensored — no API key. Point at your
// server with LLAMACPP_BASE_URL (default http://localhost:8080/v1); the
// `model` name is whatever gguf you loaded (pass-through).
Provider { key: "llamacpp", label: "llama.cpp (local)", base_url: "http://localhost:8080/v1", env_key: "LLAMACPP_API_KEY", kind: "api",
models: vec!["qwen2.5-coder-32b-instruct", "dolphin-2.9-llama3-70b", "deepseek-r1-distill-qwen-32b", "llama-3.3-70b-instruct"] },
] ]
} }
@@ -118,7 +124,7 @@ impl ChatClient {
let p = provider_for(&m.provider) let p = provider_for(&m.provider)
.ok_or_else(|| anyhow!("unknown provider '{}'", m.provider))?; .ok_or_else(|| anyhow!("unknown provider '{}'", m.provider))?;
let key = resolve_key(&p); let key = resolve_key(&p);
if key.is_empty() && p.key != "ollama" && p.key != "litellm" { if key.is_empty() && p.key != "ollama" && p.key != "litellm" && p.key != "llamacpp" {
let hint = if p.key == "gemini" { format!("{} (or GOOGLE_API_KEY)", p.env_key) } else { p.env_key.to_string() }; let hint = if p.key == "gemini" { format!("{} (or GOOGLE_API_KEY)", p.env_key) } else { p.env_key.to_string() };
return Err(anyhow!("no API key ({}) for provider '{}'", hint, p.key)); return Err(anyhow!("no API key ({}) for provider '{}'", hint, p.key));
} }
@@ -139,6 +145,7 @@ impl ChatClient {
let base = match p.key { let base = match p.key {
"litellm" => std::env::var("LITELLM_BASE_URL").unwrap_or_else(|_| p.base_url.to_string()), "litellm" => std::env::var("LITELLM_BASE_URL").unwrap_or_else(|_| p.base_url.to_string()),
"ollama" => std::env::var("OLLAMA_BASE_URL").unwrap_or_else(|_| p.base_url.to_string()), "ollama" => std::env::var("OLLAMA_BASE_URL").unwrap_or_else(|_| p.base_url.to_string()),
"llamacpp" => std::env::var("LLAMACPP_BASE_URL").unwrap_or_else(|_| p.base_url.to_string()),
_ => p.base_url.to_string(), _ => p.base_url.to_string(),
}; };
format!("{}/chat/completions", base.trim_end_matches('/')) format!("{}/chat/completions", base.trim_end_matches('/'))
@@ -178,6 +185,7 @@ impl ChatClient {
/// When `mcp_config` is set (a path to an `.mcp.json`), Claude/Codex run with /// When `mcp_config` is set (a path to an `.mcp.json`), Claude/Codex run with
/// the MCP servers enabled and tool autonomy, so agents can actually drive /// the MCP servers enabled and tool autonomy, so agents can actually drive
/// **Playwright** (browse, execute JS, screenshot) during execution. /// **Playwright** (browse, execute JS, screenshot) during execution.
#[allow(clippy::too_many_arguments)]
pub async fn chat_cli( pub async fn chat_cli(
&self, &self,
label: &str, label: &str,
+12 -10
View File
@@ -375,8 +375,8 @@ fn identify_asset(p: &crate::probe::Probe) -> String {
let tech = if p.tech.is_empty() { String::new() } else { format!(" [{}]", p.tech.join(", ")) }; let tech = if p.tech.is_empty() { String::new() } else { format!(" [{}]", p.tech.join(", ")) };
// Prefer a KNOWN product; else the org/brand from the page; else the title. // Prefer a KNOWN product; else the org/brand from the page; else the title.
let name = product let name = product
.or_else(|| if brand.is_empty() { None } else { Some(brand) }) .or(if brand.is_empty() { None } else { Some(brand) })
.or_else(|| if title.is_empty() { None } else { Some(title) }); .or(if title.is_empty() { None } else { Some(title) });
match name { match name {
Some(n) => format!("{n}{tech}"), Some(n) => format!("{n}{tech}"),
None => if tech.is_empty() { "unidentified web asset".into() } else { format!("web asset{tech}") }, None => if tech.is_empty() { "unidentified web asset".into() } else { format!("web asset{tech}") },
@@ -678,7 +678,7 @@ pub async fn run_greybox(cfg: RunConfig, lib: &Library, pool: &ModelPool, tx: Se
if !cfg.offline && !context.is_empty() { if !cfg.offline && !context.is_empty() {
let code_cap = if cfg.max_agents > 0 { cfg.max_agents.min(lib.code.len()) } else { lib.code.len().min(12) }; let code_cap = if cfg.max_agents > 0 { cfg.max_agents.min(lib.code.len()) } else { lib.code.len().min(12) };
let code_agents: Vec<Agent> = lib.code.iter().take(code_cap).cloned().collect(); let code_agents: Vec<Agent> = lib.code.iter().take(code_cap).cloned().collect();
let leads: Vec<Finding> = stream::iter(code_agents.into_iter()) let leads: Vec<Finding> = stream::iter(code_agents)
.map(|ag| { .map(|ag| {
let ctx = context.clone(); let ctx = context.clone();
let txc = tx.clone(); let txc = tx.clone();
@@ -841,7 +841,7 @@ async fn attack_chain(pool: &ModelPool, cfg: &RunConfig, recon: &str,
// Frontier = footholds to expand this round; start with confirmed, best-first. // Frontier = footholds to expand this round; start with confirmed, best-first.
let mut frontier: Vec<Finding> = confirmed.to_vec(); let mut frontier: Vec<Finding> = confirmed.to_vec();
frontier.sort_by(|a, b| sev_rank(&b.severity).cmp(&sev_rank(&a.severity))); frontier.sort_by_key(|f| std::cmp::Reverse(sev_rank(&f.severity)));
for round in 1..=max_rounds { for round in 1..=max_rounds {
if pool.stop_exploiting() || frontier.is_empty() { if pool.stop_exploiting() || frontier.is_empty() {
@@ -851,7 +851,7 @@ async fn attack_chain(pool: &ModelPool, cfg: &RunConfig, recon: &str,
let _ = tx.send(format!("⛓ attack-chain round {round}/{max_rounds} — expanding {} foothold(s), {} loot item(s)", seeds.len(), loot.len())).await; let _ = tx.send(format!("⛓ attack-chain round {round}/{max_rounds} — expanding {} foothold(s), {} loot item(s)", seeds.len(), loot.len())).await;
let loot_snapshot = loot.clone(); let loot_snapshot = loot.clone();
let results: Vec<(Vec<Finding>, Vec<String>)> = stream::iter(seeds.into_iter()) let results: Vec<(Vec<Finding>, Vec<String>)> = stream::iter(seeds)
.map(|seed| { .map(|seed| {
let (dir, rc, rb, ls, txc) = (directives.clone(), recon_ctx.clone(), recipe_block.clone(), loot_snapshot.clone(), tx.clone()); let (dir, rc, rb, ls, txc) = (directives.clone(), recon_ctx.clone(), recipe_block.clone(), loot_snapshot.clone(), tx.clone());
async move { chain_from_seed(pool, &cfg.target, &dir, &rc, &rb, &seed, &ls, round, max_rounds, &txc).await } async move { chain_from_seed(pool, &cfg.target, &dir, &rc, &rb, &seed, &ls, round, max_rounds, &txc).await }
@@ -886,7 +886,7 @@ async fn attack_chain(pool: &ModelPool, cfg: &RunConfig, recon: &str,
all_new.extend(validated.clone()); all_new.extend(validated.clone());
// Next round expands the freshly-validated footholds, best-first. // Next round expands the freshly-validated footholds, best-first.
frontier = validated; frontier = validated;
frontier.sort_by(|a, b| sev_rank(&b.severity).cmp(&sev_rank(&a.severity))); frontier.sort_by_key(|f| std::cmp::Reverse(sev_rank(&f.severity)));
} }
if !all_new.is_empty() { if !all_new.is_empty() {
let _ = tx.send(format!("⛓ attack-chaining added {} finding(s) across pivots", all_new.len())).await; let _ = tx.send(format!("⛓ attack-chaining added {} finding(s) across pivots", all_new.len())).await;
@@ -896,6 +896,7 @@ async fn attack_chain(pool: &ModelPool, cfg: &RunConfig, recon: &str,
/// Expand ONE foothold: the agent decides directions, does post-exploitation and /// Expand ONE foothold: the agent decides directions, does post-exploitation and
/// pivots, and returns new findings + discovered loot. /// pivots, and returns new findings + discovered loot.
#[allow(clippy::too_many_arguments)]
async fn chain_from_seed(pool: &ModelPool, target: &str, directives: &str, recon_ctx: &str, async fn chain_from_seed(pool: &ModelPool, target: &str, directives: &str, recon_ctx: &str,
recipe_block: &str, seed: &Finding, loot: &[String], recipe_block: &str, seed: &Finding, loot: &[String],
round: usize, max: usize, tx: &Sender<String>) -> (Vec<Finding>, Vec<String>) { round: usize, max: usize, tx: &Sender<String>) -> (Vec<Finding>, Vec<String>) {
@@ -1071,7 +1072,7 @@ fn heuristic_select(ranked: &[Agent], recon: &str, focus: &str, cap: usize) -> V
(score, a) (score, a)
}) })
.collect(); .collect();
scored.sort_by(|x, y| y.0.cmp(&x.0)); scored.sort_by_key(|x| std::cmp::Reverse(x.0));
let mut out: Vec<Agent> = scored.iter().filter(|(s, _)| *s > 0).map(|(_, a)| (*a).clone()).collect(); let mut out: Vec<Agent> = scored.iter().filter(|(s, _)| *s > 0).map(|(_, a)| (*a).clone()).collect();
if out.is_empty() { if out.is_empty() {
out = ranked.to_vec(); out = ranked.to_vec();
@@ -1082,7 +1083,7 @@ fn heuristic_select(ranked: &[Agent], recon: &str, focus: &str, cap: usize) -> V
async fn validate(candidates: Vec<Finding>, pool: &ModelPool, sys: &str, vote_n: usize, tx: &Sender<String>) -> Vec<Finding> { async fn validate(candidates: Vec<Finding>, pool: &ModelPool, sys: &str, vote_n: usize, tx: &Sender<String>) -> Vec<Finding> {
// Prefer a model other than the primary (likely finder) to adjudicate. // Prefer a model other than the primary (likely finder) to adjudicate.
let finder = pool.candidates.first().map(|m| m.label()); let finder = pool.candidates.first().map(|m| m.label());
let validated: Vec<Finding> = stream::iter(candidates.into_iter()) let validated: Vec<Finding> = stream::iter(candidates)
.map(|mut f| { .map(|mut f| {
let txc = tx.clone(); let txc = tx.clone();
let finder = finder.clone(); let finder = finder.clone();
@@ -1157,6 +1158,7 @@ async fn refute_pass(findings: Vec<Finding>, pool: &ModelPool, vote_n: usize, tx
kept kept
} }
#[allow(clippy::too_many_arguments)]
async fn finish(cfg: RunConfig, _lib: &Library, recon: String, transcript: String, mut findings: Vec<Finding>, async fn finish(cfg: RunConfig, _lib: &Library, recon: String, transcript: String, mut findings: Vec<Finding>,
selected: Vec<Agent>, rl: &mut RlState, gmode: crate::grounding::GroundMode, source_ctx: String, selected: Vec<Agent>, rl: &mut RlState, gmode: crate::grounding::GroundMode, source_ctx: String,
tx: Sender<String>) -> RunOutput { tx: Sender<String>) -> RunOutput {
@@ -1246,7 +1248,7 @@ async fn finish(cfg: RunConfig, _lib: &Library, recon: String, transcript: Strin
let mut wm = crate::belief::WorldModel::new(); let mut wm = crate::belief::WorldModel::new();
wm.deterministic = whitebox; wm.deterministic = whitebox;
for f in &findings { for f in &findings {
wm.add(&f.id, crate::belief::Kind::Exploit, &f.title, f.confidence.max(0.05).min(0.99)); wm.add(&f.id, crate::belief::Kind::Exploit, &f.title, f.confidence.clamp(0.05, 0.99));
} }
let unc = wm.uncertainty(None); let unc = wm.uncertainty(None);
if !findings.is_empty() { if !findings.is_empty() {
@@ -1275,7 +1277,7 @@ async fn finish(cfg: RunConfig, _lib: &Library, recon: String, transcript: Strin
// confirmed high-severity bugs float to the top of selection on future runs. // confirmed high-severity bugs float to the top of selection on future runs.
let mut hit: std::collections::HashMap<&str, f64> = Default::default(); let mut hit: std::collections::HashMap<&str, f64> = Default::default();
for f in &findings { for f in &findings {
let base = severity_reward(&f.severity) * f.confidence.max(0.2).min(1.0); let base = severity_reward(&f.severity) * f.confidence.clamp(0.2, 1.0);
let r = if f.review_status == "needs-review" { 0.15 } else { base }; let r = if f.review_status == "needs-review" { 0.15 } else { base };
let e = hit.entry(f.agent.as_str()).or_insert(0.0); let e = hit.entry(f.agent.as_str()).or_insert(0.0);
*e = (*e + r).clamp(-1.0, 1.0); *e = (*e + r).clamp(-1.0, 1.0);
+17 -17
View File
@@ -372,6 +372,23 @@ pub fn parse_verdict(text: &str) -> Verdict {
Verdict::Unclear Verdict::Unclear
} }
/// Severity-aware confirmation quorum. False High/Critical findings are the most
/// costly, so they require ≥2 validators AND ≥2/3 agreement; lower severities
/// pass on a strict majority (more than half). With only one validator available
/// (single-model panel) the majority rule applies to all severities.
pub fn quorum_confirmed(severity: &str, yes: usize, total: usize) -> bool {
if total == 0 {
return false;
}
let s = severity.to_lowercase();
let high = s.starts_with("crit") || s.starts_with("high");
if high && total >= 2 {
yes * 3 >= total * 2 // ≥ two-thirds
} else {
yes * 2 > total // strict majority
}
}
#[cfg(test)] #[cfg(test)]
mod verdict_tests { mod verdict_tests {
use super::*; use super::*;
@@ -405,20 +422,3 @@ mod verdict_tests {
assert!(!quorum_confirmed("Low", 0, 2)); assert!(!quorum_confirmed("Low", 0, 2));
} }
} }
/// Severity-aware confirmation quorum. False High/Critical findings are the most
/// costly, so they require ≥2 validators AND ≥2/3 agreement; lower severities
/// pass on a strict majority (more than half). With only one validator available
/// (single-model panel) the majority rule applies to all severities.
pub fn quorum_confirmed(severity: &str, yes: usize, total: usize) -> bool {
if total == 0 {
return false;
}
let s = severity.to_lowercase();
let high = s.starts_with("crit") || s.starts_with("high");
if high && total >= 2 {
yes * 3 >= total * 2 // ≥ two-thirds
} else {
yes * 2 > total // strict majority
}
}
+1 -1
View File
@@ -149,7 +149,7 @@ fn extract_brand(body: &str) -> String {
// strip the marker + a year, take the first capitalised words. // strip the marker + a year, take the first capitalised words.
let cleaned = seg.replace('©', " ").replace("&copy;", " "); let cleaned = seg.replace('©', " ").replace("&copy;", " ");
let cleaned = cleaned.trim_start_matches(|c: char| !c.is_alphabetic()); let cleaned = cleaned.trim_start_matches(|c: char| !c.is_alphabetic());
let name: String = cleaned.split(|c: char| c == '<' || c == '.' || c == '|' || c == '\n') let name: String = cleaned.split(['<', '.', '|', '\n'])
.next().unwrap_or("").chars().filter(|c| c.is_alphanumeric() || c.is_whitespace() || *c == '&' || *c == '-') .next().unwrap_or("").chars().filter(|c| c.is_alphanumeric() || c.is_whitespace() || *c == '&' || *c == '-')
.collect::<String>().trim().to_string(); .collect::<String>().trim().to_string();
// drop a leading year like "2024 " // drop a leading year like "2024 "
+3 -3
View File
@@ -232,7 +232,7 @@ fn needs_review(f: &Finding) -> bool { f.review_status == "needs-review" }
/// Strip Markdown emphasis/backticks so prose renders cleanly inside Typst. /// Strip Markdown emphasis/backticks so prose renders cleanly inside Typst.
fn strip_md(s: &str) -> String { fn strip_md(s: &str) -> String {
s.replace("**", "").replace('`', "").replace('*', "") s.replace("**", "").replace(['`', '*'], "")
} }
/// Written prose executive summary: names the asset, the counts, and the top risks. /// Written prose executive summary: names the asset, the counts, and the top risks.
@@ -313,7 +313,7 @@ pub fn markdown(target: &str, findings: &[Finding], meta: &EngagementMeta) -> St
if !meta.title.is_empty() { out.push_str(&format!("- **Page title:** {}\n", meta.title)); } if !meta.title.is_empty() { out.push_str(&format!("- **Page title:** {}\n", meta.title)); }
if !meta.tech.is_empty() { out.push_str(&format!("- **Technology:** {}\n", meta.tech.join(", "))); } if !meta.tech.is_empty() { out.push_str(&format!("- **Technology:** {}\n", meta.tech.join(", "))); }
if !meta.server.is_empty() { out.push_str(&format!("- **Server:** {}\n", meta.server)); } if !meta.server.is_empty() { out.push_str(&format!("- **Server:** {}\n", meta.server)); }
out.push_str("\n"); out.push('\n');
// --- Executive summary --- // --- Executive summary ---
out.push_str("## Executive summary\n\n"); out.push_str("## Executive summary\n\n");
@@ -333,7 +333,7 @@ pub fn markdown(target: &str, findings: &[Finding], meta: &EngagementMeta) -> St
out.push_str(&format!("| {} | {} | {} | {} | {} | {} |\n", out.push_str(&format!("| {} | {} | {} | {} | {} | {} |\n",
i + 1, f.title.replace('|', "\\|"), f.severity, owc.replace('|', "\\|"), status, auth)); i + 1, f.title.replace('|', "\\|"), f.severity, owc.replace('|', "\\|"), status, auth));
} }
out.push_str("\n"); out.push('\n');
} }
// --- Test accounts created (from the vault cleanup finding) --- // --- Test accounts created (from the vault cleanup finding) ---