agentsclimarketplace

Run3 Fuzzy Search COVERPAGE for Fund Identification

Skill cxcscmu/SkillLearnBench/skills/b3-teacher-feedback-claude-haiku-4-5/financial-analysis/run3_Fuzzy-Search-COVERPAGE-for-Fund-Identification

Search the COVERPAGE dataset using fuzzy matching to locate a specific fund by name and extract its accession_number for subsequent analysis. Use this when you need to identify a fund's filing by partial or approximate name matching.From its SKILL.md

Install
npx -y skills add cxcscmu/SkillLearnBench --skill run3_Fuzzy-Search-COVERPAGE-for-Fund-Identification

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

1.3 KB, 214 tokens by cl100k_base, as published. Nobody here has run it

Steps

  1. Load COVERPAGE data from the appropriate quarter folder (/root/2025-q2 or /root/2025-q3)

    • Read the COVERPAGE.txt file containing fund registration information
  2. Perform fuzzy matching on the fund name field

    • Use fuzzy string matching library (e.g., fuzzywuzzy) to find best match against search term
    • Set a reasonable similarity threshold (e.g., 80+) to filter candidates
    • Return top match with highest similarity score
  3. Validate the match

    • Confirm the matched fund name aligns with expected fund characteristics
    • If multiple high-confidence matches exist, manually verify which is correct
  4. Extract accession_number

    • Retrieve the accession_number field from the matched record
    • This accession_number is required for all subsequent fund detail lookups
  5. Document the match quality

    • Record the similarity score and matched fund name for audit purposes
    • Note if manual verification was required

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.