Agent Completion Enforcer
Prevents AI agents from declaring tasks complete until every requirement is verified with a one-line evidence anchor — when a gate fails, the agent stops and reports rather than proceeding, and only a human can unlock the gate to continue.
Rating
Keine Bewertungen
Sold
0
How to use
Herunterladen
Agent Completion Enforcer
The Problem
You ask your AI agent to complete a task. It says "Done."
You check. It isn't done.
The agent moved on anyway — to the next step, the next file, the next phase — carrying incomplete work forward until the error compounds into something expensive to untangle.
The problem isn't that the agent failed. The problem is that it declared success without checking.
How It Works
This skill installs a completion gate at every task boundary.
A completion gate is a set of conditions that must all be true before the agent is allowed to declare the task complete. Each condition requires a one-line evidence anchor — not a belief, not an assumption, but a concrete verifiable value.
| Without this skill | With this skill |
|---|---|
| "Done." | Gate: PASS (evidence shown) or GATE_BLOCK (missing listed) |
| Agent proceeds silently | Agent stops and reports |
| Human discovers failure later | Human decides whether to continue |
Completion Gate Format
Before declaring any task, step, or phase complete, the agent evaluates each condition:
