Gate critical AI agent actions through human approval via Telegram notifications. Developers integrate this MCP server to require manual sign-off on sensitive operations before execution.
io.github.tponscr-debug/oracle-h
Local install
STDIO
1 required env var
How models use it and what it is built for.
Gate critical AI agent actions through human approval via Telegram notifications. Developers integrate this MCP server to require manual sign-off on sensitive operations before execution.
Local install — runs as a subprocess.
Configuration this server reads at startup.
Your ORACLE-H signature key for authentication. Use 'demo-signature-key-change-in-production' for testing with no signup required.
ORACLE-H service base URL. Defaults to https://oracle-h.online
Where to find authoritative docs and source for Tponscr Debug Oracle H.
Paste any of these into Agent Studio after connecting Tponscr Debug Oracle H.
Common questions about connecting and running Tponscr Debug Oracle H.
What is oracle-h and why would I use it?
Oracle-h is an MCP server that intercepts critical AI agent actions and routes them to a human for approval via Telegram before execution. Use it when you need manual oversight on sensitive operations like deletions, config changes, or financial transactions.
How do I authenticate with oracle-h?
Set the required ORACLE_SIGNATURE_KEY environment variable with your authentication key. The default demo key is 'demo-signature-key-change-in-production'—replace it in production. Optionally configure ORACLE_BASE_URL if using a custom service endpoint (defaults to https://oracle-h.online).
How does the Telegram approval workflow work?
When your AI agent attempts a critical action, oracle-h sends a notification to your configured Telegram chat. A human reviews and approves or rejects the action, and the result is returned to the agent. The exact setup steps for linking Telegram are in the oracle-h documentation.
Can I use oracle-h for all agent actions or only specific ones?
Oracle-h is designed as a gate for critical actions—you decide which operations require approval. Common use cases are destructive operations (deletes), financial transactions, or config changes; routine reads typically bypass the gate.
What happens if a human doesn't respond to an approval request?
The MCP server documentation should specify timeout behavior (e.g., auto-reject after N minutes or wait indefinitely). Check the ORACLE_BASE_URL service docs or oracle-h-mcp package docs for timeout and retry policies.
MCP Playground runs 10,000+ hosted MCP servers — GitHub, Linear, Notion, Stripe, Sentry and more — across Claude, GPT, Gemini, DeepSeek and 60+ AI models. Compare model answers side-by-side, save agent presets, share runs. Zero install.
Open Agent StudioJudges Panel
45 judges that evaluate AI-generated code for security, cost, and quality with built-in AST.
Addozhang Nexus
Query Sonatype Nexus Repository (OSS/Pro) for Maven, Python, and Docker artifacts
Cyanheads MCP Ts Template
TypeScript template for building MCP servers with declarative tooling, observability, and auth.
Chernistry Bernstein
Declarative agent orchestration for engineering teams