feat(meta-tools): guide @mux to optimize minimal-first, expand on demand - #170
Closed
its-mash wants to merge 1 commit into
Closed
feat(meta-tools): guide @mux to optimize minimal-first, expand on demand#170its-mash wants to merge 1 commit into
its-mash wants to merge 1 commit into
Conversation
Composing a toolset was slow because the model would dump the whole catalog (mcpmux_list_all_tools), reason over all of it, then build one large FeatureSet upfront. Steer it toward a lean, incremental strategy in the guidance the model actually reads: - Server instructions: add an explicit "start minimal and expand on demand" directive — search for just the tools the task needs, compose a small FeatureSet, then grow it later (manage_feature_set update + add) only as new needs arise. Don't over-curate; re-optimizing is cheap. - mcpmux_list_all_tools description: point to mcpmux_search_tools for the common case; call out the full dump as the most token-heavy read. - mcpmux_manage_feature_set description: prefer a small initial set expanded later over adding every candidate tool upfront. Guidance-only; no behavior change to the tools themselves. Signed-off-by: Mohammod Al Amin Ashik <maa.ashik00@gmail.com>
Member
Author
|
Consolidated into #171 per request (single branch, trimmed descriptions). |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Makes
@muxtool optimization faster by steering the model toward a lean, incremental strategy instead of an exhaustive upfront one.Why
Optimizing was slow because the model would call
mcpmux_list_all_tools(dumps the entire catalog — many servers, hundreds of tools), reason over all of it, and compose one large FeatureSet in a single shot. That's token-heavy and slow, and it over-curates.Change (guidance only — no tool behavior changes)
mcp/handler.rs): add an explicit "start minimal and expand on demand" directive — search for just the tools the current task needs, compose a small FeatureSet, then grow it later (mcpmux_manage_feature_setactionupdate+add) only as new needs arise. Don't over-curate; re-optimizing is cheap.mcpmux_list_all_toolsdescription: point tomcpmux_search_toolsfor the common case and flag the full dump as the most token-heavy read.mcpmux_manage_feature_setdescription: prefer a small initial set expanded later over adding every candidate tool upfront.Verification
clippy
-D warnings, fmt, ESLint, typecheck all pass via pre-commit. No tests assert these strings (verified). The strings are model-facing guidance; effectiveness shows up as fewer/cheaper calls during@mux optimize.