Koppar

Anneal a dense placement

anneal_place

Dense placement by simulated annealing: wirelength inside the outline with courtyard gaps, edge connectors held on their edge, fixed and locked parts held, keepout rectangles kept clear (routing corridors), and a per-pad price for the bottom side (flip_cost, mm of wirelength; each bottom-side pad whose net has top-side pins costs a via). Starts from the current positions unless from_scratch. Slower than auto_place (a minute per 600k iterations) and much denser. Check overlapping_pairs and run_drc afterwards.

Input Schema

Name	Required	Description
`t0`	No	Start temperature (default 15). ~0.5 keeps the current placement and only tidies it (refine, or squeeze into a smaller outline).
`edge`	No	Connector ref to the edge it sits on, optionally @mm: where along that edge its centre goes, from the outline's top-left corner (x on top/bottom, y on left/right) — the same meaning as in auto_place and a board plan. A fixed or locked ref keeps its pin.
`halo`	No	Routing room per ref, mm (a number all round, or footprint-local [left, top, right, bottom]).
`near`	No	Keep ref at its anchor (a decoupling capacitor at its IC, a crystal at its MCU): the board plan's near constraint, as a strong pull in the anneal.
`refs`	No	The footprints that may move (default: every unlocked one).
`seed`	No
`fixed`	No	Parts held at these origin positions; they are moved there on the board first.
`shove`	No	Legaliser: a part may take a spot held only by movable parts under this fraction of its area (default 0.3, 0 off).
`gap_mm`	No
`wait_s`	No	Seconds to wait for the job (default 55). A job still running comes back with status "running": follow it with get_board_job. A job is allowed the time its budget asks for (time_budget_s, total_budget_s, rounds), and survives a server restart (the op starts over from its snapshot; route_board's pipeline goes on from its last finished stage).
`grid_mm`	No	Placement grid for free parts (0.254 = 10 mil, the designer habit); default off.
`keepouts`	No	Rectangles no part may cover, e.g. a bus corridor.
`bottom_ok`	No	Footprint-id substrings allowed on the bottom side (e.g. "0402").
`design_id`	Yes	The design id returned by import_kicad_project or list_designs.
`flip_cost`	No
`rotations`	No
`bottom_any`	No	Any SMD-only part may go to the back (priced by flip_cost); LEDs, switches, displays and test points stay on top unless bottom_exclude says otherwise.
`ic_halo_mm`	No	Room kept around ICs with 48+ pads (default 1.5).
`iterations`	No
`net_weight`	No
`refit_from`	No	Outline the current placement was drawn in; it is scaled affinely into the new outline before annealing (with t0 ~0.5: shrink a placement that closed).
`split_cost`	No	Price (mm of wirelength) of an SMD net with pads on both sides, i.e. a via.
`from_scratch`	No
`hole_room_mm`	No	Room kept around mounting holes on both sides for the screw head and standoff (default 1).
`bottom_exclude`	No	Regex of refs or footprint ids that stay on top regardless.
`overlap_weight`	No	Overlap price per mm2 at the end of the anneal (default 400; 4000 halves what a dense draw leaves).
`edge_overhang_mm`	No	Per edge ref, how far its body sits past the edge: default 0.5 for a USB/barrel/jack connector (at most its 1.2 mm lip), 0 (flush) for anything else, which may not overhang.
`connector_room_mm`	No	Room kept on the board side of every edge connector (default 2.5).
`connector_flank_mm`	No	Room kept on the flanks of every edge connector (default 1.0).
`bottom_max_height_mm`	No	Tallest part allowed on the back (3D-model height, default 1.0): the assembly box grows by the tallest back-side part.

Schema Changelog

Changes observed during successful MCP inspections.

Changed1 schema field changedOct 24, 2026

changedInput schema / properties / wait_s / description

Previous value: -"Seconds to wait for the job (default 55). A job still running comes back with status \"running\": follow it with get_board_job. A job is allowed the time its budget asks for (time_budget_s, total_budget_s, rounds), and survives a server restart (the op starts over from its snapshot)."New value: +"Seconds to wait for the job (default 55). A job still running comes back with status \"running\": follow it with get_board_job. A job is allowed the time its budget asks for (time_budget_s, total_budget_s, rounds), and survives a server restart (the op starts over from its snapshot; route_board's pipeline goes on from its last finished stage)."

Changed1 schema field changedOct 18, 2026

addedInput schema / properties / near

Added value: +{
+  "description": "Keep ref at its anchor (a decoupling capacitor at its IC, a crystal at its MCU): the board plan's near constraint, as a strong pull in the anneal.",
+  "items": {
+    "properties": {
+      "anchor": {
+        "maxLength": 40,
+        "type": "string"
+      },
+      "maxMm": {
+        "exclusiveMinimum": 0,
+        "type": "number"
+      },
+      "ref": {
+        "maxLength": 40,
+        "type": "string"
+      }
+    },
+    "required": [
+      "ref",
+      "anchor"
+    ],
+    "type": "object"
+  },
+  "maxItems": 1000,
+  "type": "array"
+}

Changed2 schema fields changedOct 18, 2026

addedInput schema / properties / fixed / description

Added value: +"Parts held at these origin positions; they are moved there on the board first."

changedInput schema / properties / wait_s / description

Previous value: -"Seconds to wait for the job (default 55). A job still running comes back with status \"running\": follow it with get_board_job."New value: +"Seconds to wait for the job (default 55). A job still running comes back with status \"running\": follow it with get_board_job. A job is allowed the time its budget asks for (time_budget_s, total_budget_s, rounds), and survives a server restart (the op starts over from its snapshot)."

Changed3 schema fields changedOct 18, 2026

changedInput schema / properties / edge / additionalProperties / pattern

Previous value: -"^(left|right|top|bottom)(@-?\\d+(\\.\\d+)?)?$"New value: +"^(left|right|top|bottom)(@\\d+(\\.\\d+)?)?$"

changedInput schema / properties / edge / description

Previous value: -"Connector ref to the edge it sits on, optionally @overhang_mm of its body past the edge (default 0.5)."New value: +"Connector ref to the edge it sits on, optionally @mm: where along that edge its centre goes, from the outline's top-left corner (x on top/bottom, y on left/right) — the same meaning as in auto_place and a board plan. A fixed or locked ref keeps its pin."

addedInput schema / properties / edge_overhang_mm

Added value: +{
+  "additionalProperties": {
+    "maximum": 5,
+    "minimum": 0,
+    "type": "number"
+  },
+  "description": "Per edge ref, how far its body sits past the edge: default 0.5 for a USB/barrel/jack connector (at most its 1.2 mm lip), 0 (flush) for anything else, which may not overhang.",
+  "propertyNames": {
+    "type": "string"
+  },
+  "type": "object"
+}

First observedOct 17, 2026

TDQS

A4.3/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Beyond the annotations (which only declare readOnlyHint=false and destructiveHint=false), the description discloses many behavioral details: it respects courtyard gaps, holds fixed/locked parts, keeps keepout rectangles clear, honors edge connector constraints, applies a per-pad price for bottom-side pads via flip_cost, starts from current positions unless from_scratch, runs at about a minute per 600k iterations, and recommends follow-up checks. This is rich context well beyond structured fields.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is front-loaded with the core purpose and is appropriately sized for a complex placement operation. The sentences carry useful information, though the first sentence is long and dense; every part still earns its place.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given a tool with 30 parameters, no output schema, and only basic safety annotations, the description provides sufficient high-level context: what it optimizes, what constraints are honored, how it starts, and what to do afterwards. It does not cover job-status semantics (covered by the wait_s schema description) or every parameter, but it is broadly complete for an agent to invoke it correctly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 77%, which is high enough that the schema itself documents most parameters. The description adds some meaning for a few parameters (e.g., flip_cost as 'mm of wirelength', edge connectors held on their edge, fixed/locked parts held, keepouts kept clear), but it does not explain the majority of the 30 parameters. Baseline 3 is appropriate.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states a specific verb and resource: 'Dense placement by simulated annealing'. It also distinguishes the tool from its sibling auto_place by noting it is 'slower than auto_place ... and much denser', so an agent can tell when this tool is the right choice.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides clear context for use: it is slower but much denser than auto_place, starts from current positions unless from_scratch, and advises to 'Check overlapping_pairs and run_drc afterwards'. It does not explicitly state when to avoid this tool or provide a full decision tree, but the alternative and trade-off are clear.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All Koppar tools Inspect

Try in Browser

Claude Code, inside the Godot 4 editorFlockbay puts an AI agent in the real Godot 4 editor: it writes GDScript, runs your game and fixes what breaks. Free, or bring your own Claude Code or Codex.flockbay.comAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources

Need Help?