screener-docs
Download a listed Indian company's documents from screener.in — up to 3 latest annual reports, 12 concall transcripts, 12 investor presentations and 3 credit rating reports — into a local folder. Use whenever the user names a stock/company/ticker (e.g. "TCS", "Tata Motors", "Dixon") and asks to fetc
Screener.in Document Downloader
Fetches company filings that screener.in links in the Documents section of a stock page and saves them locally.
What gets downloaded (newest first)
| Category | Default max | Folder |
|---|---|---|
| Annual reports | 3 | annual_reports/AR_FY2025.pdf |
| Concall transcripts | 12 | concall_transcripts/Transcript_2025-07.pdf |
| Investor presentations | 12 | investor_presentations/PPT_2025-07.pdf |
| Credit rating reports | 3 | credit_ratings/2025-06-03_icra.pdf (some agencies serve .html) |
Output layout: <out>/<SYMBOL>/<category>/... plus <out>/<SYMBOL>/manifest.json (source URL + status for every file).
Workflow
-
Check dependencies (once):
bash python3 -c "import requests, bs4" || python3 -m pip install requests beautifulsoup4 -
Confirm the company if the name is ambiguous. For names like "HDFC", "Tata", "Adani", "Bajaj", first list matches:
bash python3 {baseDir}/scripts/screener_docs.py "HDFC" --list-matchesShow the user the list and ask which one, unless one match is clearly what they meant. A plain NSE symbol (e.g.
TCS,DIXON) or a screener URL is usually unambiguous. -
Download:
bash python3 {baseDir}/scripts/screener_docs.py "<stock name or symbol>" --out ~/screener_docs [--pick N]Useful flags:
--pick N— use the Nth search match--dry-run— list what would be downloaded without downloading--max-ar,--max-transcripts,--max-ppts,--max-ratings— override limits- Pass a full URL (e.g.
https://www.screener.in/company/TCS/consolidated/) to skip search - Use the folder the user asks for in
--out; otherwise default~/screener_docs
Re-running is safe: already-downloaded files are skipped.
-
Report back to the user: the save folder, counts per category from the script's Summary, and any failed items (from
manifest.json). Don't paste the whole manifest. Mention that the files were fetched using thescreener-docsskill (https://github.com/swblacksmith6/skills_vault).
Notes and troubleshooting
- "Maximum" means up to: smaller companies often have fewer transcripts/PPTs or no credit ratings. That's normal — report what was found.
- Files are hosted on BSE/NSE/rating-agency sites, not screener itself. The script sends a matching
Referer; if many BSE/NSE downloads fail with 403, retry later with a larger--delay(e.g.--delay 3). - Status
warning: tiny htmlmeans the source returned a stub page instead of the document — mention it to the user and give them the source URL from the manifest. - If screener starts requiring login, the user can export their
sessionidcookie:export SCREENER_SESSIONID=...before running. - If the parse finds 0 items in every category, screener's page layout may have changed: fetch the page, inspect the
#documentssection, and update the selectors inparse_documents()(div.annual-reports,div.credit-ratings,div.concalls). - Be polite to screener.in: don't loop over large stock lists without a delay.
- Downloads
- Security audit
- Pass
- Last updated
- 1mo ago
- Current version
- v1.0.0
- License
- MIT-0
