Design Observer · An explainable lens on design

# Core capabilities

# 1. Summarize

Generate structured summaries for any snapshot: cell/net statistics, critical-path delay distribution, congestion hotspot coordinates, power heatmaps, clock-tree depth and skew. Agents get a high-level design overview without parsing the raw database. Summary output is structured JSON for LLM context or Model Router input features.

# 2. Diff

Compare structured differences between two snapshots. Output three change sets: which cells changed (position, size, type), which nets were rerouted (topology, layer), and how timing changed (critical-path slack delta, new and resolved violations). Diff results feed Agent Planner as a what-changed signal.

# 3. Explain

Explain the causal chain of a change. For example: why did resizing this cell slow that path? Observer traces capacitance change → drive strength → transition time → cell delay, outputting readable causal steps. Each step includes quantitative data—not just slower, but +42ps because fanout cap rose from 0.08pF to 0.15pF.

# 4. Root Cause

Trace violations to root cause. Given a DRC or timing violation, Observer answers: which change introduced it, which Agent proposal caused it, and possible fix paths. Chain: violation → diff → change → candidate → agent intention.

# Agent invocation

Observer exposes a unified API for all four primitives; Agents switch modes via the observe_type parameter.

from ieda.observer import DesignObserver

obs = DesignObserver("http://localhost:9110")

# Summarize: get design overview
summary = obs.observe(
    design_ref="snap_b",
    observe_type="summarize",
)
print(summary.cell_count)     # 45231
print(summary.critical_paths) # [{"name":"clk_core","slack":-0.12},...]

# Diff: compare two snapshots
diff = obs.observe(
    design_ref="snap_b",
    observe_type="diff",
    reference_ref="snap_a",
)

# Root cause: trace violation to agent proposal
rc = obs.observe(
    design_ref="snap_b",
    observe_type="root_cause",
    violation_ref="DRC_M1_SPACING_x1245_y330",
)
print(rc.introducing_change)  # "cell_move_u_alu_0"
print(rc.agent_proposal)      # "candidate_c42_a"

// Summarize request
{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "observer.observe",
    "arguments": {
      "design_ref": "snap_b",
      "observe_type": "summarize"
    }
  }
}

// Diff request
{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "observer.observe",
    "arguments": {
      "design_ref": "snap_b",
      "observe_type": "diff",
      "reference_ref": "snap_a"
    }
  }
}

# Summarize design state
observer observe -design snap_b -type summarize

# Diff two snapshots
observer observe -design snap_b -type diff -reference snap_a

# Explain a change
observer observe -design snap_b -type explain \
  -change cell_resize_u7

# Root cause analysis
observer observe -design snap_b -type root_cause \
  -violation DRC_M1_SPACING_x1245_y330

# Input/output contract

ObserveRequest

ObserveResult

# Related resources