- Home
- Skills
- Agents & Orchestration
- From a Sentence to a Verified Robot Mission
From a Sentence to a Verified Robot Mission
Plain words to a mission that was simulated and measured: quantitative criteria agreed first, a gate that refuses what was not tested.
$49
Works with the AI Agents you already use
From a Sentence to a Verified Robot Mission
Example session with this skill installed
Plan a drone inspection for the north face of the dam. We need at least 98% coverage and must finish with 20% battery. Use Gazebo for the run.
- Read your context and instructions
- Compiled the from a sentence
I have extracted the mission intent and generated the quantitative specification. The simulation run has been completed and verified against the acceptance gate.
Mission Verification Report
| Metric | Threshold | Measured | Result |
|---|---|---|---|
couverture_pct | >= 98% | 99.2% | PASS |
collisions | == 0 | 0 | PASS |
marge_batterie_pct | >= 20% | 24.5% | PASS |
duree_min | <= 60 | 42.0 | PASS |
Simulation Metadata
- Simulator: Gazebo 11
- Seed: 20261021
- Status: All blocking criteria satisfied.
Mission Plan Specification
{
"mission_id": "dam-north-face-001",
"criteria": [
{"metrique": "couverture_pct", "operateur": ">=", "seuil": 98, "unite": "%"},
{"metrique": "collisions", "operateur": "==", "seuil": 0},
{"metrique": "marge_batterie_pct", "operateur": ">=", "seuil": 20, "unite": "%"}
],
"plan": {
"pattern": "lawnmower_vertical",
"step_over": 1.2,
"altitude_offset": 5.0,
"waypoints": [
{"lat": 45.123, "lon": 5.456, "alt": 210},
{"lat": 45.124, "lon": 5.456, "alt": 180}
]
}
}
Next steps
- Review the
plan.jsonwaypoints for local no-fly zone compliance. - Export the mission package to the drone's ground control station.
- Perform a manual pre-flight check on the real asset to account for wind conditions not modeled in Gazebo.
Connects securely to your tools. The creator never sees your data.
About this skill
Deliver only what passed criteria agreed before the run.
Turn a sentence into a robot mission that has been simulated and measured before anyone sees it - extract the intent into a specification with quantitative acceptance criteria, plan, run it in simulation, compare the measurements to the criteria agreed beforehand, and deliver only what passed. Use when a user describes a mission in plain words, when generated robot code must be proven before deployment, or when a team needs a gate between a simulation run and a real robot.
Generating a plan from "inspect the south face of the bridge" is the easy part, and it is the part everyone ships. The hard part - and the only one worth selling - is refusing to hand it over until a run has produced numbers that meet criteria written down beforehand. "Good coverage" is not a criterion; coverage at or above 95 % is.
Tested scripts: scripts/porte_acceptation.py compares the run's measurements to the agreed criteria and exits non-zero unless every blocking criterion passes. It also refuses four things that look like success: a criterion that was agreed and never measured, because silence is not a pass; a measurement with no criterion, a number nobody agreed to judge; a criterion that is not quantitative, rejected when the file is read; and a run with zero coverage or no recorded random seed, because a verdict nobody can reproduce is an opinion with a number attached. Seven cases verified, including a passing run and each refusal.
Includes an EXPERT.md reference (exact parameters, thresholds, failure diagnosis): how to write criteria that mean something and why every recall threshold needs a precision one beside it, the reasoning behind each refusal, reproducibility requirements, a diagnosis table mapping each gate failure to what to look at first, the two extra criteria a multi-robot mission needs, and a table separating what a passing simulation establishes from what it does not.
That last table is the honest part: a passing simulation says the plan is coherent and nothing collides in the model. It says nothing about wind on the day, GPS multipath under a deck, or the real texture of corroded steel. The skill delivers that boundary with the mission.
Acceptable use: civil and commercial missions only - inspection, mapping, logistics, agriculture, maintenance, research, education. The skill refuses any military, paramilitary or weapon-related use, targeting or surveillance of persons, declaring a mission verified when no run took place, and lowering a criterion to make a run pass without the user deciding it explicitly.
It will not fly anything, will not claim a simulation it did not run, and will not turn a passing simulation into a guarantee about a real asset.
Requires a local simulation environment (Gazebo/Isaac Sim). Does not account for real-world environmental noise like wind or GPS multipath. Prohibits all military or surveillance applications.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 27 days ago
Requires