lean-ctx 3.6.5

Context Runtime for AI Agents with CCP. 51 MCP tools, 10 read modes, 60+ compression patterns, cross-session memory (CCP), persistent AI knowledge with temporal facts + contradiction detection, multi-agent context sharing, LITM-aware positioning, AAAK compact format, adaptive compression with Thompson Sampling bandits. Supports 24+ AI tools. Reduces LLM token consumption by up to 99%.
Documentation
//! Maps lean-ctx security features to the OWASP Top 10 for Agentic Applications (2025).
//! Used by `lean-ctx audit` CLI and `/v1/audit/events` endpoint.

use serde::Serialize;

#[derive(Debug, Clone, Serialize)]
pub struct OwaspMapping {
    pub owasp_id: &'static str,
    pub owasp_title: &'static str,
    pub risk_description: &'static str,
    pub lean_ctx_mitigations: Vec<Mitigation>,
    pub coverage: Coverage,
}

#[derive(Debug, Clone, Serialize)]
pub struct Mitigation {
    pub feature: &'static str,
    pub module: &'static str,
    pub description: &'static str,
}

#[derive(Debug, Clone, Copy, Serialize, PartialEq)]
pub enum Coverage {
    Full,
    Partial,
    Minimal,
}

pub fn alignment() -> Vec<OwaspMapping> {
    vec![
        OwaspMapping {
            owasp_id: "OWASP-AGENT-01",
            owasp_title: "Excessive Agency",
            risk_description: "Agent performs actions beyond intended scope or without proper authorization",
            lean_ctx_mitigations: vec![
                Mitigation { feature: "Capability System", module: "core/capabilities.rs", description: "Fine-grained capability declarations per tool (fs:read, fs:write, exec, net)" },
                Mitigation { feature: "Role Guard", module: "server/role_guard.rs", description: "5 built-in roles with tool allowlists and shell policy" },
                Mitigation { feature: "Shell Allowlist", module: "core/shell_allowlist.rs", description: "Opt-in command allowlist restricting which binaries agents can execute" },
                Mitigation { feature: "Context Budget", module: "core/agent_budget.rs", description: "Per-agent token budgets preventing resource exhaustion" },
            ],
            coverage: Coverage::Full,
        },
        OwaspMapping {
            owasp_id: "OWASP-AGENT-02",
            owasp_title: "Prompt Injection",
            risk_description: "Malicious instructions injected via data processed by the agent",
            lean_ctx_mitigations: vec![
                Mitigation { feature: "Context Compression", module: "core/terse/", description: "Deterministic compression reduces attack surface in injected content" },
                Mitigation { feature: "Secret Detection", module: "core/secret_detection.rs", description: "Pre-read scanning detects and optionally redacts sensitive patterns" },
                Mitigation { feature: "I/O Boundary", module: "core/io_boundary.rs", description: "Content filtering and secret-like path blocking before agent consumption" },
            ],
            coverage: Coverage::Partial,
        },
        OwaspMapping {
            owasp_id: "OWASP-AGENT-03",
            owasp_title: "Sensitive Information Disclosure",
            risk_description: "Agent exposes confidential data through outputs or tool interactions",
            lean_ctx_mitigations: vec![
                Mitigation { feature: "PathJail", module: "core/pathjail.rs", description: "Filesystem jail prevents reads outside project root" },
                Mitigation { feature: "I/O Boundary", module: "core/io_boundary.rs", description: "Secret-like path detection (.env, .ssh, credentials)" },
                Mitigation { feature: "Secret Detection", module: "core/secret_detection.rs", description: "Regex-based detection of API keys, tokens, passwords in file content" },
                Mitigation { feature: "Proxy Header Allowlist", module: "proxy/forward.rs", description: "Prevents leaking Set-Cookie and other sensitive headers" },
                Mitigation { feature: "Memory Boundary", module: "core/memory_boundary.rs", description: "Cross-project access control with audit trail" },
            ],
            coverage: Coverage::Full,
        },
        OwaspMapping {
            owasp_id: "OWASP-AGENT-04",
            owasp_title: "Denial of Service",
            risk_description: "Agent overwhelms system resources or causes service disruption",
            lean_ctx_mitigations: vec![
                Mitigation { feature: "Rate Limiter", module: "core/a2a/rate_limiter.rs", description: "Per-agent per-tool rate limiting" },
                Mitigation { feature: "Memory Guard", module: "core/config/memory.rs", description: "RAM usage caps and idle cleanup" },
                Mitigation { feature: "Budget Tracker", module: "core/agent_budget.rs", description: "Token budget enforcement with hard limits" },
                Mitigation { feature: "Loop Detection", module: "config loop_detection", description: "Detects and throttles repetitive tool call patterns" },
                Mitigation { feature: "Tool Timeout", module: "engine/mod.rs", description: "120s timeout on tool execution prevents indefinite hangs" },
            ],
            coverage: Coverage::Full,
        },
        OwaspMapping {
            owasp_id: "OWASP-AGENT-05",
            owasp_title: "Supply Chain Vulnerabilities",
            risk_description: "Compromised tools, plugins, or dependencies affect agent behavior",
            lean_ctx_mitigations: vec![
                Mitigation { feature: "Signed Handoff Bundles", module: "core/handoff_transfer_bundle.rs", description: "Ed25519 signatures verify integrity and provenance of transferred data" },
                Mitigation { feature: "Audit Trail", module: "core/audit_trail.rs", description: "SHA-256 chained append-only log of all tool calls and security events" },
                Mitigation { feature: "Agent Identity", module: "core/agent_identity.rs", description: "Per-agent Ed25519 keypairs for cryptographic identity" },
            ],
            coverage: Coverage::Partial,
        },
        OwaspMapping {
            owasp_id: "OWASP-AGENT-06",
            owasp_title: "Insufficient Logging and Monitoring",
            risk_description: "Lack of visibility into agent actions and security events",
            lean_ctx_mitigations: vec![
                Mitigation { feature: "Audit Trail", module: "core/audit_trail.rs", description: "Every tool call logged with agent ID, role, input hash, output tokens" },
                Mitigation { feature: "Compliance Reports", module: "cli/audit_report.rs", description: "CLI command to generate aggregated compliance reports" },
                Mitigation { feature: "Context OS Events", module: "core/context_os.rs", description: "Real-time event bus with SSE streaming for dashboard" },
                Mitigation { feature: "Proxy Metrics", module: "proxy/metrics.rs", description: "Atomic counters for requests, tokens saved, bytes compressed" },
            ],
            coverage: Coverage::Full,
        },
        OwaspMapping {
            owasp_id: "OWASP-AGENT-07",
            owasp_title: "Insecure Code Execution",
            risk_description: "Agent executes arbitrary or unsafe code without proper sandboxing",
            lean_ctx_mitigations: vec![
                Mitigation { feature: "Sandbox Level 0", module: "core/sandbox.rs", description: "Subprocess isolation with env_clear and timeout" },
                Mitigation { feature: "Sandbox Level 1 (macOS)", module: "core/sandbox_seatbelt.rs", description: "OS-level Seatbelt profiles restricting filesystem and network" },
                Mitigation { feature: "Sandbox Level 1 (Linux)", module: "core/sandbox_landlock.rs", description: "Landlock LSM restricting filesystem access" },
                Mitigation { feature: "Command Blocklist", module: "tools ctx_shell", description: "Dangerous command patterns blocked before execution" },
            ],
            coverage: Coverage::Full,
        },
        OwaspMapping {
            owasp_id: "OWASP-AGENT-08",
            owasp_title: "Broken Access Control",
            risk_description: "Agent accesses resources or performs actions beyond its permissions",
            lean_ctx_mitigations: vec![
                Mitigation { feature: "RBAC", module: "core/roles.rs", description: "5 built-in roles (viewer, coder, admin, ci, restricted) with granular policies" },
                Mitigation { feature: "Capability System", module: "core/capabilities.rs", description: "Tool-level capability requirements checked against role grants" },
                Mitigation { feature: "PathJail", module: "core/pathjail.rs", description: "All path arguments jailed to project root" },
                Mitigation { feature: "Boundary Policy", module: "core/memory_boundary.rs", description: "Cross-project access control configurable per policy" },
            ],
            coverage: Coverage::Full,
        },
        OwaspMapping {
            owasp_id: "OWASP-AGENT-09",
            owasp_title: "Improper Multi-Agent Orchestration",
            risk_description: "Coordination failures between agents lead to conflicts or data corruption",
            lean_ctx_mitigations: vec![
                Mitigation { feature: "Per-Agent Ledger", module: "core/context_ledger.rs", description: "Isolated context tracking per agent, preventing cross-contamination" },
                Mitigation { feature: "Agent Registry", module: "core/agents.rs", description: "HTTP-backed registration with heartbeat and auto-deregistration" },
                Mitigation { feature: "TaskStore File Locks", module: "core/a2a/task.rs", description: "Advisory file locks prevent lost updates from concurrent access" },
                Mitigation { feature: "Atomic Writes", module: "core/context_ledger.rs", description: "Crash-safe temp+rename writes for all JSON stores" },
            ],
            coverage: Coverage::Full,
        },
        OwaspMapping {
            owasp_id: "OWASP-AGENT-10",
            owasp_title: "Insufficient Governance",
            risk_description: "Lack of organizational policies and controls over agent behavior",
            lean_ctx_mitigations: vec![
                Mitigation { feature: "Policy Engine", module: "core/context_policies.rs", description: "Declarative policies with agent, content, and time-based conditions" },
                Mitigation { feature: "Compliance Reports", module: "cli/audit_report.rs", description: "Aggregated reports: reads, compressions, denials, budget usage" },
                Mitigation { feature: "Auto-Reroot Protection", module: "tools/server_paths.rs", description: "Opt-in control over project root changes, audited" },
                Mitigation { feature: "Config-Driven", module: "core/config/mod.rs", description: "All security features configurable via config.toml" },
            ],
            coverage: Coverage::Full,
        },
    ]
}

/// Returns a compact summary suitable for CLI output.
pub fn summary() -> String {
    let mappings = alignment();
    let mut out = String::from("OWASP Top 10 for Agentic Applications — lean-ctx Alignment\n");
    out.push_str(&"=".repeat(60));
    out.push('\n');
    for m in &mappings {
        let icon = match m.coverage {
            Coverage::Full => "",
            Coverage::Partial => "",
            Coverage::Minimal => "",
        };
        out.push_str(&format!(
            "\n{icon} {}{}\n  Mitigations: {}\n",
            m.owasp_id,
            m.owasp_title,
            m.lean_ctx_mitigations
                .iter()
                .map(|m| m.feature)
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    let full = mappings
        .iter()
        .filter(|m| m.coverage == Coverage::Full)
        .count();
    let partial = mappings
        .iter()
        .filter(|m| m.coverage == Coverage::Partial)
        .count();
    out.push_str(&format!(
        "\nCoverage: {full}/10 Full, {partial}/10 Partial\n"
    ));
    out
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn alignment_covers_all_ten() {
        let a = alignment();
        assert_eq!(a.len(), 10);
        for (i, m) in a.iter().enumerate() {
            assert_eq!(m.owasp_id, format!("OWASP-AGENT-{:02}", i + 1));
            assert!(!m.lean_ctx_mitigations.is_empty());
        }
    }

    #[test]
    fn summary_contains_all_ids() {
        let s = summary();
        for i in 1..=10 {
            assert!(s.contains(&format!("OWASP-AGENT-{i:02}")));
        }
    }
}