
@black_wall
Building Black_Wall — a pre-action risk gate for AI agents. It forecasts what an action will do and returns GO / CONFIRM / STOP before the agent deletes prod, drops a table, or sends money. API + MCP, free tier. blackwalltier.com
Building Black_Wall — a pre-action risk gate for AI agents. It forecasts what an action will do and returns GO / CONFIRM / STOP before the agent deletes prod, drops a table, or sends money. API + MCP, free tier. blackwalltier.com
What did I ship? The distribution layer.
Black_Wall — a pre-action risk gate for AI agents (GO / CONFIRM / STOP before it does something irreversible) — is now one config line away in basically any agent stack:
→ npm: npx blackwall-mcp — runs in Claude Desktop/Code, Cursor, Windsurf, Antigravity
→ Listed in the official MCP registry
→ Drop-in adapters: LangChain, CrewAI, Vercel AI SDK, OpenAI tool calling
→ Plain HTTP API + OpenAPI for everything else
Write the gate once, reach every host. (Glama + awesome-mcp-servers listings in review.)
blackwalltier.com
Building Black_Wall — a pre-action risk gate for AI agents. It forecasts what an action will do and returns GO / CONFIRM / STOP before the agent deletes prod, drops a table, or sends money. API + MCP, free tier. blackwalltier.com
First post here. Building Black_Wall — a pre-action risk gate for AI agents (one call before an irreversible action → GO / CONFIRM / STOP).
Just shipped /stress-test: list the actions your agent can take and it runs each through 28 documented failure modes, then hands back a report card of which ones it'd wave through that it shouldn't. Built it because every agent disaster this year was an action nothing checked first.
blackwalltier.com/stress-test