BLACK_WALL

A pre-action risk gate for autonomous AI agents. Before an agent takes an irreversible or high-stakes action, it calls one tool and gets back a risk score, a reversibility class, named red flags, and a verdict — in a few seconds — so it knows whether to proceed, confirm with a human, or stop.

Free tier, no credit card. Get a key at https://blackwalltier.com

What it does

It runs before an action, not after. It's the judgment layer that decides whether a proposed action should happen now — catching what static permission rules can't:

• Prompt injection
• Anomalous amounts
• PII in the wrong place
• Destructive SQL (DELETE/DROP with no WHERE)
• Prod-from-staging
• Irreversible deletes with no backup

It complements rollback, backups, audit logs, and observability — all of which act after the damage. Black_Wall acts before.

The one tool: forecast

Call it with the action and its payload:

• action — what the agent intends to do (send_email, run_sql, make_payment, file_delete, post_content, transfer_crypto...). Free-form is fine.
• inputs — the concrete payload (recipient, amount, SQL statement, file path, message body).
• context (optional) — agent_role, user_intent, prior_actions, prior_findings.

You get back:

• recommendation — GO / CAUTION / STOP
• risk_score — 0–100
• reversibility — REVERSIBLE / RECOVERABLE / IRREVERSIBLE (+ rollback cost & window)
• gate — AUTO / CONFIRM / HUMAN_REQUIRED (derived from risk AND reversibility — an irreversible action is gated at a lower threshold)
• red_flags — named codes with severity
• alternative_actions — concrete safer actions to take instead

How an agent acts on the verdict

• AUTO / GO → proceed.
• CONFIRM / CAUTION → confirm with a human first.
• HUMAN_REQUIRED / STOP → do not act. The action has not executed — don't build downstream logic on it. Surface the red flags and pivot to an alternative.

Worked example

Request: action=run_sql, inputs={"statement":"DELETE FROM users;"}

Response: recommendation=STOP, risk_score=92, reversibility=IRREVERSIBLE, gate=HUMAN_REQUIRED, red_flags=[SQL_NO_WHERE, DESTRUCTIVE_VERB, IRREVERSIBLE_NO_BACKUP], alternative_actions=["Add a WHERE clause scoped to inactive customers and take a snapshot first"].

Install

{
  "mcpServers": {
    "blackwall": {
      "command": "npx",
      "args": ["-y", "blackwall-mcp"],
      "env": { "BLACKWALL_API_KEY": "bw_live_your_key_here" }
    }
  }
}