Which AI agents are compatible with CodeStory?

CodeStory is designed for any Python-capable AI agent; it provides a command-line interface and a structured JSON schema that agents use to interpret execution traces and compile them into visual stories.

What is included in the CodeStory package?

You receive the core Python tracer, the storyboard generator agent, the custom HTML5 animation engine, and full documentation on the structured storyboard schema.

Does CodeStory work with my existing data stack?

CodeStory is built for Python 3.8+ and includes specialized support for tracing libraries like pandas, requests, and common SQL drivers like psycopg2 and sqlite3.

Why should I use this instead of just asking an LLM to explain my code?

Unlike generic LLM summaries, CodeStory uses a dedicated animation engine and actual telemetry data to ensure every row count and data transformation is grounded in fact rather than model hallucination.

Works with the AI Agents you already use

+22 more

+18 agents

codestory

Q: How does the 'Safe-mode' tracing protect my production environment?

The 'Safe-mode' feature automatically intercepts and redirects side effects like file writes, database mutations, and network requests to a sandbox, allowing you to trace production scripts without affecting live data.

Q: Are there any limits on how much code can be visualized at once?

To keep stories concise and readable, the tool automatically truncates traces for functions called more than three times under the same parent, preventing repetitive loops from bloating the visualization.

Turn Python scripts and execution traces into interactive, animated HTML stories for stakeholders and engineers.

Free

Security scanned

codestory

Example session with this skill installed

/codestory run process_daily.py --audience stakeholders -- --date 2024-05-20

Read your context and instructions
Compiled the codestory
Generated the data export

I have recorded the execution of process_daily.py for the requested date. Safe mode was active: 4 CSV writes were redirected to the sandbox, and 1 Slack notification was blocked.

Daily Processing Story: May 20, 2024

The execution successfully processed the daily weather telemetry. 842 raw sensor readings were validated and aggregated into a single summary report.

{
  "title": "Daily Weather Pipeline",
  "subtitle": "Processes raw sensor telemetry into daily aggregates.",
  "entry": "process_daily.py --date 2024-05-20",
  "status": "success",
  "scenes": [
    {
      "title": "Load raw telemetry",
      "actor": {
        "label": "load_data()",
        "sublabel": "io_utils.py",
        "code": {
          "file": "io_utils.py",
          "line": 12,
          "snippet": "def load_data(path):\n    df = pd.read_csv(path)\n    return df",
          "highlight": [13]
        }
      },
      "outputs": [
        {
          "kind": "data",
          "label": "raw_df",
          "rows": 842,
          "cols": 6,
          "sample": [["2024-05-20 00:01", "STN_01", 22.4]]
        }
      ],
      "narration": "The script loaded 842 readings from the daily CSV export. All columns, including temperature and humidity, were parsed correctly."
    },
    {
      "title": "Filter invalid readings",
      "actor": {
        "label": "clean_readings()",
        "sublabel": "filters.py",
        "code": {
          "file": "filters.py",
          "line": 45,
          "snippet": "df = df[df['temp'].between(-50, 60)]\nreturn df",
          "highlight": [45]
        }
      },
      "inputs": [{ "kind": "data", "label": "raw_df", "rows": 842 }],
      "outputs": [{ "kind": "data", "label": "clean_df", "rows": 838 }],
      "stats": [
        { "label": "Removed", "value": 4, "delta": -4 }
      ],
      "narration": "Four readings were dropped because the temperature values were outside the sensor's physical operating range (-50 to 60°C)."
    }
  ]
}

Next steps

Open story.html in any browser to view the animated walkthrough.
Inspect the 4 sandboxed CSV files in /tmp/codestory/run_8f2d/.
Run with --live if you want to commit the summary to the production database.

story.html

CSV · data export

Generated

Example file from a real run - the skill writes it into your workspace.

Connects securely to your tools. The creator never sees your data.

About this skill

Turn complex Python codebases or specific execution runs into interactive, animated HTML stories. This skill bridges the gap between raw source code and stakeholder comprehension by generating a visual "storyboard" that maps data flow, transformations, and decisions. It uses a custom Python tracer to record exactly what happened during a run—including data samples, row counts, and intercepted side effects—or scans the logic to explain how a system works without executing it.

The tool includes a built-in animation engine (player.html) that produces a single, offline-ready file. It handles "safe mode" execution, automatically sandboxing file writes, database mutations, and network requests so you can trace production scripts locally without side effects.

What it does

Visualizes execution by generating an animated timeline of function calls, data states, and I/O events.
Records data transformations with automatic tracking of table shapes (rows/cols) and data samples.
Safe-mode tracing intercepts and redirects side effects to a temporary sandbox for risk-free analysis of live scripts.
Audience-tuned narration generates plain-language explanations of code logic for stakeholders or deep-dives for engineers.
Branch analysis identifies taken and untaken paths in a specific run to debug logic flow.

How it works

Record: Run codestory.py scan to map logic or run to trace a live execution with specific arguments.
Storyboard: The agent analyzes the resulting JSON bundle to draft a structured storyboard.json following the official schema.
Build: Run the build command to embed the storyboard into a standalone, animated story.html player.

Frameworks & tools

Built for Python 3.8+ using the standard library. Supports tracing for pandas, requests, httpx, psycopg2, pymysql, and sqlite3.

Why this beats prompting it yourself

Generic LLMs struggle to maintain consistent line-number highlights and arithmetic accuracy across thousands of lines of trace data. This skill uses a structured schema and a dedicated animation engine to ensure every data card, code snippet, and row count is grounded in actual execution telemetry rather than hallucinated summaries.

Use cases

Onboarding engineers by providing an animated walkthrough of a complex pipeline's entry point and data flow.
Incident post-mortems to visualize exactly why a specific run failed or where data was dropped.
Stakeholder reporting to show non-technical managers how a business process translates into code logic.
Safe refactoring by comparing "scan" stories of new logic against "run" stories of legacy scripts.

Known limitations

Does not support cloud SDKs (AWS/GCP), MongoDB, or Redis interception. Limited to Python. Truncates detailed traces for functions called more than three times under the same parent.

How to use

Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.

Reviews

No reviews yet

Be one of the first to try it. Every listed skill passes our trust checks below.

Security scanned

Passed our 8-point scan before listing

Fresh listing

Recently published to Agensi

Free forever

No account required to browse

Trust & safety

Security scanned

Verified clean 10 days ago

Free to download with an account

Requires

Googleapis

Gstatic

Python

Listed10 days ago

codestory