elastic/sysgrok

Issues

• analyzecmd: Final output from multi-stage summarisation should read as if the summary was computed without intermediate summarisation

  #21 opened a year ago by SeanHeelan
  0

• findfaster: Update the description of requirements for alternatives

  #20 opened a year ago by SeanHeelan
  0

• findfaster: Add requirement that any suggestions must not use the target library as a dependency

  #19 opened a year ago by SeanHeelan
  0

• cmdanalysis: Experiment with detecting/including a header in data for chunk summaries

  #18 opened a year ago by SeanHeelan
  0

• cmdanalysis: Validate whether or not chunking of command output loses information

  #17 opened a year ago by SeanHeelan
  0

• debughost: Revisit how commands are generated, executed, and checked for correctness

  #16 opened a year ago by SeanHeelan
  0

• Tracker: Improvements to command output analysis #6

  #1 opened a year ago by SeanHeelan
  0

• debughost: Handle syntax errors in suggested commands

  #7 opened a year ago by SeanHeelan
  1

• debughost: LLM may suggest commands that do not terminate

  #9 opened a year ago by SeanHeelan
  1

• debughost: Allow user to specify a file containing a list of commands to run

  #15 opened a year ago by SeanHeelan
  0

• debughost: Does forcing the LLM to provide an explanation for why it is suggesting each command lead to more useful commands

  #14 opened a year ago by SeanHeelan
  0

• Binary distribution for perf-copilot

  #13 opened a year ago by iogbole
  0

• cmdanalysis: Need to implement chunked handling of the output of commands which may exceed the token limit

  #12 opened a year ago by SeanHeelan
  0

• Command analysis can result in output which refers to the example

  #10 opened a year ago by SeanHeelan
  0

• debughost: LLM does not differentiate between commands that should be run on the host being debugged versus on another host

  #8 opened a year ago by SeanHeelan
  0

• debughost: Need to handle cases where command output is too long for prompt

  #6 opened a year ago by SeanHeelan
  0

• topn: LLM not identifying all possible problems given list of Python functions

  #5 opened a year ago by SeanHeelan
  0

• Fix TopN command so that optimisations are tied to the input data

  #4 opened a year ago by SeanHeelan
  1

• Add functionality to count tokens in queries + responses and log them for debugging purposes

  #3 opened a year ago by SeanHeelan
  0

• Investigate whether the multiple prompts in findfaster can be integrated into one

  #2 opened a year ago by SeanHeelan
  0