An M3 tool to teach how to create new M3 Tool: Early stage #52
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Outdated suggestions cannot be applied.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.
Hiya folks! Promise it won't be as long as #51.
The following is a first attempt at educating users / contributors to construct M3 ready-tools based on a user query. The tool essentially teaches the LLM the structure required, best practices, and so on through a small number of MCP actions so that it does not stray too much, as we need to lean a little bit in the deterministic sphere for such a process (i.e. Any incoming tools should strictly be following the M3's architecture).
Below see a quick successful video where I build first an M3's pipeline with
MIMIC
and theM3ToolBuilder
to challenge theLLM
with the creation of — for the sake of the example because it was the simplest example that came to me — the MIT beaver mascot M3 tool (PS: one prompt was enough for the tool to be working with M3 and Claude MCP host), allowing users to know a bit about it even though it's a pretty-bad example given that few LLM web-searches suffices. However, I would appreciate it if we could all contribute with the following in the comments:<your proposed prompt>
.# Focus M3's scope proposed tools
I will try them all, one by one, and we will basically analyse the tool's viability as such. If we can verify tools from
EASY
toCOMPLEX
prompt without having to re-prompt 50 times for it to be working, I believe theM3ToolBuilder
tool might be a useful educational resource for contributors.Caution
Yes, this PR is based on #51; whenever #51 sees the light of day, I'll rebase here, leaving this PR with only two commits (as of now). In the meantime, see more at for the specific commits #52 comes with: See changes
Important
This is still in its early stages, so I suppose any ideas are welcome as well as prompts to challenge the tool.
M3ToolBuilderCompressed.mp4
Cheers,