Musmem Locations

• "IFBB, 2025" → fetch listing page for IFBB in 2025, extract date+location for all contests
• "2025" → fetch listing pages for all orgs (IFBB, NPC, NPC Worldwide, CPA) for 2025
• "NPC, 2023-2025" → fetch listing pages for NPC for each year in the range

Workflow: fetch npcnewsonline.com listing pages → parse date+location in bulk → match to MuscleMemory contest names → write results (see Path B Workflow below).

Python scripting rule: Never use python3 -c "..." or python3 - <<'PYEOF' heredocs. Always write scripts to /tmp/script.py using the Write tool, then run python3 /tmp/script.py.

MuscleMemory API

GET https://musclememory.net/api/contests                          # all contests across all orgs
GET https://musclememory.net/api/contests/{year}                   # all contests across all orgs for a year
GET https://musclememory.net/api/org?name={ORG}                    # all contest names for one org (use to find exact contest name)
GET https://musclememory.net/api/contest/years?name={contest}      # all years DB has results for a specific contest

Use browser User-Agent for all musclememory.net API calls (server blocks bots):

Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36

JSON Schema

File: ~/workspace/musmem/contest_locations.json

[
  {
    "contest": "Contest Name - ORG",
    "years": {
      "1940": {
        "date": "May 25, 1940",
        "venue": "Madison Square Garden",
        "location": "New York, New York, USA"
      },
      "1947": { "date": "", "venue": "", "location": "" }
    }
  }
]

• date — free text, can be a range (e.g., "June 1-2, 1946"), empty string if unknown
• venue — building/arena name, empty string if unknown
• location — "City, State, Country" format, empty string if unknown
• Write valid JSON after every update

Path B Workflow

Step 1 — Fetch listing pages

Use curl + Python (write script to /tmp/). Listing URLs from npcnewsonline.com:

User-Agent for curl: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36

Step 2 — Parse date and location

Write a Python script to /tmp/parse_listing.py that parses the HTML and extracts, for each contest:

• Contest name (as it appears on the site — will be normalized in next step)
• Date
• Location / venue (if present on the listing page)

Examine the HTML structure on first use to identify the relevant tags and classes.

Step 3 — Normalize contest names

Apply the same normalization as musmem-contests:

• Strip the leading org prefix and space (e.g., IFBB Arnold Classic → Arnold Classic)
• Append - {ORG} (e.g., Arnold Classic - IFBB)

Step 4 — Match to MuscleMemory and write

• For each parsed contest, check contest_locations.json for an existing entry
• Skip years that already have data (unless the user gave a specific year)
• Write date, venue, and location fields for each matched contest+year
• Write valid JSON after every update

Step 5 — Report

List contests written, skipped (existing data), and any that could not be matched to a MuscleMemory contest name.

Path A Workflow

Step 1 — Resolve contest name

If the user gave the exact contest name (e.g., "Nationals - NPC"), use it directly.

If the name is ambiguous or the org is given instead:

• GET /api/org?name={ORG} → pick the matching contest name from the list

Step 2 — Determine target years

Specific year given (e.g., "Nationals - NPC, 1995"):

• That single year is the only target. Proceed regardless of existing JSON data.

Year range given (e.g., "1982-2025" or "1982 to present"):

• GET /api/contest/years?name={contest} → get years in DB
• Filter to the requested range
• Skip any year that already has ANY field filled in the JSON
• Remaining years are the targets

No year given:

• GET /api/contest/years?name={contest} → get all years in DB
• Skip any year that already has ANY field filled in the JSON
• Remaining years (all fields empty, or year not yet in JSON) are the targets

Step 3 — Research each target year

Process sequentially — one year at a time. Use Search Tiers below.

Step 4 — Write results

Specific year was given AND data already exists in JSON:

• If found data differs from existing: present both versions, ask for confirmation before updating
• If found data matches existing: report "no change"
• If fields were empty: write directly

All other cases:

• Write found data directly to JSON after each year (don't batch at the end)
• If the year doesn't exist in JSON yet, create the entry

Step 5 — Report

After all targets: list what was found, what was skipped (had existing data), and what remains empty (not found after exhausting all search tiers).

Search Tiers

Tier 1 — Org website (modern contests):

• For IFBB/NPC/NPC Worldwide/CPA: check npcnewsonline.com contest listing or result page — often includes date and venue
• If not found: search for promoter's event page: "[Contest Name] [year] [org]" schedule OR results

Tier 2 — General web search:

• "[Contest Name] [year]" bodybuilding date location
• "[Contest Name] [year]" bodybuilding "[org]"
• For historical contests (pre-1980): also try Wikipedia and history sites like sandowplus.co.uk

Tier 3 — Declare unknown:

• After 2–3 failed searches across both tiers, leave fields empty and note "not found" in report

Common Mistakes