- Home
- Skills
- Agents & Orchestration
- Agent Action Proof — Verify MCP Writes
Works with the AI Agents you already use
Agent Action Proof — Verify MCP Writes
Preflight one authorized MCP tool write, verify its intended final state by authoritative readback, and resolve timeouts before retrying.
Free
Agent Action Proof — Verify MCP Writes
Example session with this skill installed
Illustrative fictional request: “I authorize moving task T-42 in workspace North to Ready. Check its current status, use the connected task tool once if needed, then read the task back. If the tool times out, check whether the change landed before retrying.”
- Read your context and instructions
- Compiled the agent action proof
Illustrative output only; no task provider was contacted.
Target: task T-42 in workspace North. Requested postcondition: status equals Ready. Preflight: confirm the account and task identity from an authoritative read, inspect the current update-tool schema, and record the task's current status and revision. If the status is already Ready, report “already satisfied” and do not write. Otherwise make one authorized update and retain the provider's request ID.
The completion receipt should include the write response, the exact readback source and time, and the observed status. If the readback proves T-42 is Ready, label the action completed. If a documented operation result conclusively says no change took effect, label it not completed. If the call timed out and the readback could be stale, label outcome unknown and name the next safe check. Do not issue a second update merely because the first response is missing. A retry is allowed only after verified idempotency or conclusive no-effect evidence, within the original authorization. This sample demonstrates the decision format; it is not evidence that T-42 or a provider exists.
Connects securely to your tools. The creator never sees your data.
About this skill
An agent changes a task, issue, document, or other connected resource. The tool says success—or times out. Did the intended change actually happen?
Agent Action Proof checks one authorized external write at the point of action. It identifies the exact target, current tool schema, baseline state, and observable postcondition before the call. Afterward it reads the target system and returns a short receipt: completed, not completed, or outcome unknown. An ambiguous response triggers reconciliation before any retry.
Use it to verify an MCP tool action, confirm a connected-app write, or decide whether a timed-out action can be retried safely. It handles one mutation boundary, not a multi-step automation design.
How to use
Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scanned
Verified clean 5 days ago
- Free to download with an account