# io.github.Al-Adnane/kasbah-mcp (npm · @yobekasbah/mcp-server)

Governance layer for agentic AI — signed, verifiable receipts for every agent action.

- Trust score: 61/100 (medium)
- Change this week: −10
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-08-03

## Components

- npm · `@yobekasbah/mcp-server`: 61/100 (this document), [markdown](https://verifymcp.io/servers/al-adnane-kasbah-mcp/yobekasbah-mcp-server.md), [page](https://verifymcp.io/servers/al-adnane-kasbah-mcp/yobekasbah-mcp-server)

## Channel facts

- Registry: `npm`
- Package: `@yobekasbah/mcp-server`
- Version: `3.0.1`
- Transport: `stdio`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog)

Scored 2026-08-03.

- **Supply Chain Security**: 100/100
  - No malware found by supply-chain analysis.
  - No known CVEs affecting this package version or its production dependencies.
  - No install/post-install scripts declared.
  - No production dependencies, so there is no dependency health to assess.
- **Provenance & Transparency**: 45/100
  - Source repository is publicly reachable at the declared URL.
  - Provenance check failed: no build-provenance attestation is published.
  - Clear OSI-approved license (MIT).
  - Actively maintained (last published 27 days ago).
  - Disclosure check failed: no security disclosure policy was found in the source repository.
- **Schema Quality & AI Usability**: 42/100
  - AI-judged instruction clarity (poor).
  - Context-footprint check failed: tool/resource definitions use about 1767 tokens (~110/item across 16 items; 16 tools + 0 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 27/100
  - Stability observed for 8 of 30 days with no destabilising changes; credit accrues until the full window elapses.
- **Tool Coverage**: 90/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 69% of tool parameters carry a description.
- **Capabilities**: 20/100
  - Spec-recency check failed: implements MCP spec 2024-11-05; the latest is 2026-07-28.

## Install

### Claude

```bash
claude mcp add al-adnane-kasbah-mcp -- npx -y @yobekasbah/mcp-server
```

### Codex

```bash
codex mcp add al-adnane-kasbah-mcp -- npx -y @yobekasbah/mcp-server
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "al-adnane-kasbah-mcp": {
      "type": "local",
      "command": [
        "npx",
        "-y",
        "@yobekasbah/mcp-server"
      ],
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add al-adnane-kasbah-mcp --command npx --arg -y --arg @yobekasbah/mcp-server
```

### Hermes

```yaml
mcp_servers:
  al-adnane-kasbah-mcp:
    command: "npx"
    args: ["-y", "@yobekasbah/mcp-server"]
```

### Other

```json
{
  "mcpServers": {
    "al-adnane-kasbah-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@yobekasbah/mcp-server"
      ]
    }
  }
}
```

## Changelog

Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims.

### 2026-08-02 (score 61, +19)

- [security improvement] Malware scan: unverified → pass
- [functional improvement] Stability: unverified → 0.23

### 2026-08-01 (score 42, −11)

- [functional] We updated how we score, so this day's move reflects our rubric, not a change to the server

### 2026-07-31 (score 53, −18)

- [security regression] Malware scan: pass → unverified

### 2026-07-27 (score 71, 0)

- [functional] We updated how we score, so this day's move reflects our rubric, not a change to the server

### 2026-07-26 (score 71)

First indexed and scored.

## MCP tools (16)

### `kasbah_attest` (~180 tokens)

Call this BEFORE any side-effecting action (file write, shell exec, HTTP request, payment, tool call, agent spawn). Returns a verdict (ALLOW / WARN / DENY) and a cryptographic receipt anyone can verify offline, forever. If the verdict is DENY, do not proceed.

Input parameters:

- `action` (string, required): short verb for what you are about to do (e.g. "file_write", "shell_exec", "http_post", "agent_spawn")
- `context` (string): optional: why you are doing this (for the auditor)
- `input` (string, required): the actual content you are about to act on — command, file body, URL, prompt. Hashed before signing.
- `surface` (string): optional: where you run (e.g. "claude-code", "cursor", "agent-runtime")

### `kasbah_verify` (~130 tokens)

Verify a Kasbah receipt cryptographically. Supports v2 (Ed25519) and v3 (Ed25519 + Merkle proofs). Returns whether the signature is genuine, the verdict, when it was signed, and by which engine key. Works offline.

Input parameters:

- `merkleProof` (array): optional Merkle proof siblings (v3)
- `merkleRoot` (string): optional Merkle root for batch verification (v3)
- `payload` (string, required): base64url-encoded canonical JSON payload
- `receipt` (string, required): kasbah_receipt:v3:ed25519:…

### `kasbah_status` (~57 tokens)

Show what Kasbah has done for you: decisions signed, actions blocked, $ saved from cheaper-model routing, top token sinks, receipts minted, and a link to your live dashboard. Real value-for-money at a glance. No arguments.

### `kasbah_export_audit` (~101 tokens)

Export a tamper-evident, auditor-ready bundle of recent signed decisions. Each entry carries the verdict, Ed25519 signature, public-key fingerprint, and spec version. Hand it to SOC 2 / EU AI Act auditors — they verify offline.

Input parameters:

- `limit` (number): how many recent decisions to export (default 100, max 1000)
- `verdict` (string): optional filter: ALLOW | WARN | DENY

### `kasbah_check_compliance` (~276 tokens)

Evaluate an agent action against pre-certified HIPAA, GDPR, SOC 2 Type II, PCI-DSS, and BIOMETRIC_GUARD rule packs BEFORE it runs. BIOMETRIC_GUARD blocks facial micro-expression / silent-speech / subvocal / BCI / QAI surveillance attempts. Returns ALLOW / WARN / DENY / REQUIRE_APPROVAL with the cited rule ID (e.g. "H-002: SSN requires redaction", "B-001: facial micro-expression capture prohibited"). Use for any action touching regulated data (PHI, PII, payment data) or biometric/intent-inference surfaces.

Input parameters:

- `content` (string): optional payload to scan for regulated data (SSN, card numbers, PHI) or surveillance vectors (silent-speech, subvocal, faceprint)
- `packs` (string): comma-separated packs (default "HIPAA,GDPR,SOC2,BIOMETRIC_GUARD"). Options: HIPAA, GDPR, SOC2, PCI_DSS, BIOMETRIC_GUARD
- `target` (string): target resource (e.g. patient_records, payment_data, user_profile, facial_expression)
- `verb` (string, required): action verb: READ / WRITE / DELETE / EXPORT / EXEC / CAPTURE_VIDEO / ACCESS_CAMERA …

### `kasbah_trace_intent` (~107 tokens)

Detect distributed, multi-step attacks invisible to single-step scanners: data-exfiltration pipelines, privilege-escalation chains, cost bombs. Pass the recent action steps (one "VERB target" per line). Returns CLEAN / WARN / THREAT / CRITICAL with the matched attack trajectory and confidence.

Input parameters:

- `steps` (string, required): recent steps, one per line as "VERB target" (e.g. "READ patient_db\nENCODE base64\nSEND external.com")

### `kasbah_cost_guard` (~142 tokens)

Runaway-loop and spend governor. Checks whether an agent is stuck in a tool loop or burning budget, and recommends ALLOW / THROTTLE / BLOCK before the next call runs. Returns the decision plus live usage (cost/hr, tokens/min, brittleness band). Call before expensive or repeated tool calls.

Input parameters:

- `agent` (string, required): agent id (e.g. "orchestrator")
- `calls` (number): how many identical recent calls have already happened (loop signal)
- `tokens` (number): tokens per call (default 1200)
- `tool` (string, required): tool about to be called (e.g. "web_search")

### `kasbah_mcp_servers` (~34 tokens)

List all MCP servers governed by Kasbah (ensemble, openclaw, ansari, quranfrontier).

### `kasbah_mcp_teams` (~88 tokens)

Manage governance teams. Create, add members, assign MCPs, apply HIPAA/GDPR/SOC2/PCI-DSS policies.

Input parameters:

- `action` (string, required)
- `mcpServer` (string)
- `passportId` (string)
- `policyId` (string)
- `teamId` (string)
- `teamName` (string)

### `kasbah_govern_turn` (~161 tokens)

Govern a full LLM turn through the Kasbah Governance Proxy: pass the message history and any proposed tool calls; get a bounded advisory verdict (CLEAR / WARN / REQUIRE_APPROVAL) plus an authoritative DENY for forbidden verbs / non-whitelisted domains / rate limits. Runs the validated detector, fail-closed governance engine, conformal-calibrated advisory ensemble, multi-step attack-trajectory detection, and a forward-secure audit. Use to wrap any agent turn before it executes tools.

Input parameters:

- `clearance` (number): caller clearance 0-4 (OBSERVER..DIRECTOR), default 2
- `messages` (array, required): conversation messages [{role, content}]
- `tools` (array): proposed tool calls [{verb, target, domain}]

### `kasbah_advisory` (~141 tokens)

Run a single advisory analyzer directly. Modules: detectgpt (AI-text), intent (multi-step trajectory), chsh, conformal (anytime-valid), dobft (perturbed-observer collusion), flag-fusion (correlation-aware evidence), evasion-bound (orthogonal-manifold cost), bounded (advisory operator). All are bounded-advisory and never DENY.

Input parameters:

- `module` (string, required)
- `payload` (object): module-specific body (e.g. {text} for detectgpt, {steps} for intent, {score} for conformal/bounded, {flags} for flag-fusion)

### `kasbah_os` (~91 tokens)

Access the kasbah-os engine (Python bridge). Defensive actions are live: health, govern (9-module RKHS ensemble turn), audit-verify. The offensive Mythos modules are introspection-only and cannot be executed.

Input parameters:

- `action` (string, required)
- `clearance` (number)
- `messages` (array): for action=govern: [{role, content}]
- `tools` (array)

### `kasbah_scan` (~56 tokens)

Quick content scan through the Kasbah detector. Returns CLEAR / WARN / REQUIRE_APPROVAL with threat classes and risk score. Use for single-prompt or short content safety checks.

Input parameters:

- `text` (string, required): content or prompt to scan

### `kasbah_redteam` (~57 tokens)

Run the live red-team benchmark against the current governance pipeline. Returns detection metrics (F1, precision, recall), degradation alerts, and trend data. Use to verify the system is working right now.

Input parameters:

- `action` (string, required)

### `kasbah_fti_query` (~82 tokens)

Query the Federated Threat Intelligence registry. Checks whether a prompt hash matches any known attack in the shared Merkle tree registry. Privacy-preserving: only SHA-256 hashes are submitted, never raw prompts.

Input parameters:

- `hash` (string): SHA-256 hash of prompt to query
- `text` (string): raw prompt (hashed locally, never transmitted)

### `kasbah_compliance_report` (~64 tokens)

Generate SOC 2 Type II / ISO 27001:2022 / NIST CSF 2.0 compliance evidence packets from the audit chain. Returns JSON or print-ready HTML. Use for auditor requests or compliance automation.

Input parameters:

- `framework` (string, required)

## Diagnostics

Captured diagnostic sections: Provenance, Dependencies. The full working is on the page: https://verifymcp.io/servers/al-adnane-kasbah-mcp/yobekasbah-mcp-server#diagnostics

## Score history

- 2026-08-03: 61
- 2026-08-02: 61
- 2026-08-01: 42
- 2026-07-31: 53
- 2026-07-29: 71
- 2026-07-28: 71
- 2026-07-27: 71
- 2026-07-26: 71

## Links

- npm package: https://www.npmjs.com/package/@yobekasbah/mcp-server
- Socket report: https://socket.dev/npm/package/@yobekasbah/mcp-server
- Repository: https://github.com/Al-Adnane/Kasbah-Core
- Changelog RSS feed: https://verifymcp.io/servers/al-adnane-kasbah-mcp/yobekasbah-mcp-server/changelog.xml
- Changelog JSON feed: https://verifymcp.io/servers/al-adnane-kasbah-mcp/yobekasbah-mcp-server/changelog.json
- HTML version of this page: https://verifymcp.io/servers/al-adnane-kasbah-mcp/yobekasbah-mcp-server