vedic-chart - SKILL.md Agent Skill

name: vedic-chart description: This skill should be used when the user shares a Vedic/Jyotish horoscope chart image and asks to "convert this chart", "extract this kundli", "digitize this horoscope", "chart to ascii", "read this birth chart", "what planets are in this chart", or to parse a rasi/navamsa/divisional (D-1, D-9, D-7, D-30, etc.) chart printout. Converts a chart image into structured JSON plus ASCII North-Indian and South-Indian diagrams. Also invoked by the download:blog-post skill when astrology charts are found in an article.

Vedic Chart Extractor

Convert a Vedic astrology chart image (a Jagannatha Hora / Parashara's Light style printout) into a structured JSON chart and ASCII North-Indian + South-Indian diagrams. A bundled script does the work end to end.

What the source images look like

The script targets the standard software printout: a North-Indian chart (left), a South-Indian chart (right), a birth-data header, and — crucially — a longitude table across the bottom listing each body as sign deg' min with retrograde and Jaimini karaka markers (e.g. Su: 4 Sc 08-DK). That bottom table is the ground truth; houses are computed from it deterministically, so the chart diagrams are never parsed glyph-by-glyph.

A chart-only image with no bottom table still works, but bodies the table would have carried fall back to sign-only placement (no degrees).

How it works

image --(claude vision reads the longitude table)--> chart JSON --(pure renderers)--> ASCII

House placement is house = (sign_index - lagna_index) mod 12 + 1. North-Indian is house-fixed (Lagna always top-center, signs rotate); South-Indian is sign-fixed (Pisces/Aries/… in fixed cells, houses rotate). Both render from the same JSON.

Usage

Extraction calls the claude CLI (must be on PATH, recent enough to accept --effort and --allowedTools). Run the bundled script:

# extract + render both ASCII styles; JSON then charts to stdout
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/chart_to_ascii.py path/to/chart.png

# write sidecars (<stem>.json + <stem>.txt) next to the image(s)
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/chart_to_ascii.py images/*.png -o images/

# render only, no vision call, from a previously extracted JSON
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/chart_to_ascii.py chart.json --emit ascii

Options: --emit {json,ascii,both} (default both), --style {north,south,both} (default both), -o/--outdir DIR, --model MODEL (default claude-opus-4-8), --overwrite (re-extract even if a sibling .json exists; otherwise a cached sidecar is reused). Data goes to stdout, progress/errors to stderr, non-zero exit on any failure.

The script is standalone (plain python3, no virtualenv or extra packages).

Procedure

Confirm the input is a Vedic chart image (or a chart .json for render-only).
Run the script with -o <dir> to write .json + .txt sidecars next to the image, or without -o to print to stdout for a quick look.
For batches, pass a glob; the script processes each file and reuses cached .json sidecars unless --overwrite is given.
Surface the ASCII chart and note the sidecar paths. Do not hand-transcribe the chart — the script's JSON is the structured source of truth.

JSON shape

{
  "chart_type": "Rasi",
  "native": "Indira Gandhi",
  "date": "1917-11-19", "time": "23:11:00", "tz": "5:30 east",
  "place": "81 E 51, 25 N 27",
  "lagna": "Cn",
  "bodies": [
    {"code": "As", "sign": "Cn", "deg": 27, "minute": 23},
    {"code": "Su", "sign": "Sc", "deg": 4, "minute": 8, "karaka": "DK"},
    {"code": "Ju", "sign": "Ta", "deg": 14, "minute": 6, "retro": true, "karaka": "MK"}
  ]
}

code uses two-letter abbreviations: planets/points As Su Mo Ma Me Ju Ve Sa Ra Ke and upagrahas AL HL GL SL Md Gk. sign is Ar Ta Ge Cn Le Vi Li Sc Sg Cp Aq Pi. retro and karaka (Jaimini chara karaka: AK AmK BK MK PiK PK GK DK) appear only when present.