mdicio commited on
Commit
85c5434
·
1 Parent(s): a2d8a72

requirements

Browse files
Files changed (1) hide show
  1. agent.py +7 -7
agent.py CHANGED
@@ -213,7 +213,7 @@ class BoomBot:
213
  • Do NOT make unfounded assumptions—always ground answers in reliable sources or search results.
214
  • For math or puzzles: break the problem into code/math, then solve programmatically.
215
 
216
- RESEARCH WORKFLOW (in rough priority order):
217
  1. SEARCH
218
  - Try web_search, wikipedia_search, or arxiv_search first.
219
  - Refine your query rather than repeating the exact same terms.
@@ -227,13 +227,13 @@ class BoomBot:
227
  4. DOWNLOAD
228
  - Use download_file_from_link tool on relevant links found (yes you can download webpages as html).
229
  - For arXiv papers, target the /pdf/ or DOI link (e.g https://arxiv.org/pdf/2011.10672).
230
- -
231
  5. INDEX & QUERY
232
  - Add downloaded documents to the vector store with add_document_to_vector_store.
233
  - Use query_downloaded_documents for detailed answers.
234
  6. READ
235
- - You have access to a read_file_content tool to read most types of files. You can also directly interact with downloaded files in your python code (do this for csv files and excel files)
236
-
 
237
 
238
  FALLBACK & ADAPTATION:
239
  • If a tool fails, reformulate your query or try a different search method before dropping to download.
@@ -261,9 +261,9 @@ class BoomBot:
261
 
262
  FINAL ANSWER FORMAT:
263
  - Begin with "FINAL ANSWER: "
264
- - Number → digits only (e.g., 42)
265
- - String → exact text (e.g., Pope Francis)
266
- - List → comma-separated, one space (e.g., 2, 3, 4)
267
  - Conclude with: FINAL ANSWER: <your_answer>
268
  """
269
 
 
213
  • Do NOT make unfounded assumptions—always ground answers in reliable sources or search results.
214
  • For math or puzzles: break the problem into code/math, then solve programmatically.
215
 
216
+ RESEARCH WORKFLOW:
217
  1. SEARCH
218
  - Try web_search, wikipedia_search, or arxiv_search first.
219
  - Refine your query rather than repeating the exact same terms.
 
227
  4. DOWNLOAD
228
  - Use download_file_from_link tool on relevant links found (yes you can download webpages as html).
229
  - For arXiv papers, target the /pdf/ or DOI link (e.g https://arxiv.org/pdf/2011.10672).
 
230
  5. INDEX & QUERY
231
  - Add downloaded documents to the vector store with add_document_to_vector_store.
232
  - Use query_downloaded_documents for detailed answers.
233
  6. READ
234
+ - You have access to a read_file_content tool to read most types of files (html, pdf, text).
235
+ - You can also directly interact with downloaded files (csv, excel) in your python code.
236
+ - Use query_downloaded_documents if you have added docs to vector store.
237
 
238
  FALLBACK & ADAPTATION:
239
  • If a tool fails, reformulate your query or try a different search method before dropping to download.
 
261
 
262
  FINAL ANSWER FORMAT:
263
  - Begin with "FINAL ANSWER: "
264
+ - Number → digits only (e.g., 42) no units unless specified
265
+ - String → exact text (e.g., Pope Francis) without quotation marks
266
+ - List → comma-separated, one space, no brackets unless specified(e.g., 2, 3, 4)
267
  - Conclude with: FINAL ANSWER: <your_answer>
268
  """
269