| name | code-visualization |
| description | Use when a request asks to visualize code, repository structure, runtime behavior, state, data movement, dependencies, types, proof state, an interactive graph, or a diagram embedded in a document; infer the user's context question, embedding context, and precision need, select the diagram family and evidence source, then delegate rendering to the owning skill or tool. |
Code Visualization
Reader Map
- Purpose: runtime skill for selecting the right diagram or visualization route
for code, repository structure, runtime behavior, proof state, or data flow.
- Use When: a task asks to visualize code, dependencies, state transitions,
architecture, proofs, or document-embedded diagrams.
- Tool Commands: run this skill's command packet, then read the canonical
agents/skills/code-visualization.md selection rules.
- Boundary: visualization must be based on the owning evidence source; diagrams
do not replace implementation, proof, or dependency checks.
Tool Commands
Use the command packet before applying this skill's workflow:
python3 tools/agent_tools/skill_tool_commands.py show --skill code-visualization --format text
Execute the required and task-matching conditional commands that the packet prints.
- Read
agents/skills/code-visualization.md.
- Record a context-derived
Visualization Selection before rendering:
context_question
embedding_context
precision_need
visualization_kind
question
source_evidence
owner_skill_or_tool
renderer
output_path
- Infer the context question, then project it to a diagram family:
- "what happens in what order": flowchart / activity diagram.
- "which exact branches and joins exist": control-flow graph.
- "what calls or imports what": call graph or dependency graph.
- "who exchanges messages over time": sequence diagram.
- "how concurrent events overlap": timing diagram or concurrency sequence diagram.
- "what states can exist and how transitions occur": state-transition diagram.
- "where data or artifacts move": data-flow diagram.
- "which types, classes, protocols, or owners relate": class/type diagram or
architecture map.
- "where algorithm/proof status sits on implemented operations":
$algorithm-flowchart.
- "which large graph needs filtering or navigation":
$html-output after the
graph source is available.
- For a diagram embedded in a document, infer the local claim, section role,
reader action, and
visual_plan slot before choosing the diagram family.
Pair this skill with $structure-planning for the visual plan and
$md-style-check for Mermaid / Markdown checks.
Treat this as Document Embedded Diagrams: the section claim, reader path,
and embedding context are part of the visualization selection.
- Route source evidence through the owning tool or skill:
- dependency graph:
bash tools/agent_tools/check_dependency_graph.sh --changed --print-edges
- file-level code dependency surface:
bash tools/agent_tools/scan_code_dependencies.sh --changed
- Python function call surface:
python3 tools/agent_tools/helper_function_inventory.py --changed --all-functions --format json
- skill and owner selection:
python3 tools/agent_tools/route.py --prompt "<user request>" --format json
- related skill command packets:
python3 tools/agent_tools/skill_tool_commands.py show --skill dependency-analysis --format text
python3 tools/agent_tools/skill_tool_commands.py show --skill structure-planning --format text
python3 tools/agent_tools/skill_tool_commands.py show --skill algorithm-flowchart --format text
python3 tools/agent_tools/skill_tool_commands.py show --skill structure-refactor --format text
python3 tools/agent_tools/skill_tool_commands.py show --skill prose-reasoning-graph --format text
python3 tools/agent_tools/skill_tool_commands.py show --skill html-output --format text
python3 tools/agent_tools/skill_tool_commands.py show --skill md-style-check --format text
- Keep pass/fail authority with the source producer. The diagram is a
projection of extracted facts; code, dependency, proof, or runtime checkers
own correctness claims.