Describe the feature or problem you'd like to solve
Provide local session logs (similar to e.g. Claude Code) for sessions
Proposed solution
All of the other major AI development tools (Claude, Codex, Gemini) store machine-readable logs of the users' sessions, with a bunch of specific information about what happened during a user's session (you can find these in e.g. ~/.claude/projects/*/*.jsonl). copilot-cli doesn't seem to create these (just an unformatted debug log), but it should!
Having these logs means that external tools can be used to examine sessions and understand the nitty-gritty of what's actually happening under the hood in those sessions -- quite useful for figuring out why the AI tooling is behaving certain ways, what it is doing in terms of tool use, where during a session tokens are being used (and how many), what specific files are getting read at what point, etc. See e.g. https://claude-dev.tools/ for one of my favorites.
Without these logs being available, this kind of tooling is just completely unavailable for copilot. This type of tool seems quite popular in the larger community (because it's a great way to study and improve one's workflows!) so not having those logs (and associated tooling being able to consume them) is potentially a barrier to adoption.
Example prompts or workflows
No response
Additional context
No response
Describe the feature or problem you'd like to solve
Provide local session logs (similar to e.g. Claude Code) for sessions
Proposed solution
All of the other major AI development tools (Claude, Codex, Gemini) store machine-readable logs of the users' sessions, with a bunch of specific information about what happened during a user's session (you can find these in e.g.
~/.claude/projects/*/*.jsonl). copilot-cli doesn't seem to create these (just an unformatted debug log), but it should!Having these logs means that external tools can be used to examine sessions and understand the nitty-gritty of what's actually happening under the hood in those sessions -- quite useful for figuring out why the AI tooling is behaving certain ways, what it is doing in terms of tool use, where during a session tokens are being used (and how many), what specific files are getting read at what point, etc. See e.g. https://claude-dev.tools/ for one of my favorites.
Without these logs being available, this kind of tooling is just completely unavailable for copilot. This type of tool seems quite popular in the larger community (because it's a great way to study and improve one's workflows!) so not having those logs (and associated tooling being able to consume them) is potentially a barrier to adoption.
Example prompts or workflows
No response
Additional context
No response