feat(mcp): discovery/validate/dry-run/run meta-tools + Glama metadata - #44
Merged
Merged
Conversation
…data The MCP server previously exposed only one tool per workflow file, so an empty directory produced zero tools (nothing to introspect or grade) and an agent could run a workflow but not discover, validate, or preview one. - Add four always-present, well-described meta-tools with correct MCP annotations: list_workflows, validate_workflow, dry_run_workflow (all read-only) and run_workflow (destructive). tools/list is now non-empty even with an empty directory. - Fix a latent bug: per-workflow execution called engine.run(**arguments), but run()'s signature is run(params=...); parameterized workflows would have raised TypeError. Now passes params correctly. - Capture engine stdout (e.g. the dry-run preview) so it cannot corrupt the stdio JSON-RPC stream; return it as 'preview' for dry runs. - Keep the per-workflow convenience tools (additive). - Add glama.json (maintainer: orieg) for Glama server verification. - Docs: document the tool model; tests for resolve/validate/execute helpers. Verified end-to-end over MCP stdio: 4 meta-tools list with an empty dir; run_workflow and per-workflow tools execute with params.
There was a problem hiding this comment.
Pull request overview
This PR expands the MCP server surface so agents can discover, validate, preview (dry-run), and run workflows via four always-present “meta-tools”, while preserving the existing per-workflow convenience tools. It also adds Glama metadata and updates docs/tests to reflect and verify the new tool model.
Changes:
- Add four meta-tools (
list_workflows,validate_workflow,dry_run_workflow,run_workflow) with MCP annotations and richer descriptions. - Fix workflow execution parameter passing (engine
run(params=...)) and capture stdout to protect the JSON-RPC stream (returning dry-run output aspreview). - Add
glama.json, document the tool model, and add tests for workflow resolution/validation/execution helpers.
Reviewed changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
src/yaml_workflow/mcp_server.py |
Introduces meta-tools, workflow resolution/validation/execute helpers, annotations, and updated dispatch logic. |
tests/test_mcp_server.py |
Adds unit tests for the new helper functions and dry-run preview behavior. |
docs/guide/mcp.md |
Updates documentation to describe the meta-tools + per-workflow tool model and MCP annotations. |
glama.json |
Adds Glama server metadata for verification/scoring. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
- list_workflows description now says it returns an object with {count,
workflows} (matching the actual return), not an array
- dry_run_workflow no longer claims 'no files are written': it does not
execute tasks or cause the workflow's side effects, but writes ephemeral
logs to a temporary workspace — description + docs corrected to match the
read-only annotation without contradiction
Owner
Author
|
Addressed all three in df6ecb4: |
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
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.
Summary
Makes the MCP server a first-class agent surface — and prepares it for a Glama A-grade (the awesome-mcp-servers badge gate) by ensuring
tools/listis non-empty and every tool is well-described with correct annotations.Previously the server exposed only one tool per workflow file, so:
Changes
Four always-present meta-tools with rich, Glama-graded descriptions and correct MCP annotations:
list_workflowsvalidate_workflowdry_run_workflowrun_workflowDescriptions and
readOnlyHint/destructiveHintagree (no contradictions — a Glama scoring axis).Latent bug fixed: per-workflow execution called
engine.run(**arguments), but the signature isrun(params=...)— any parameterized workflow would have raisedTypeError. Now passes params correctly.stdio safety: engine stdout (e.g. the dry-run preview) is captured so it can't corrupt the JSON-RPC stream; returned as
previewfor dry runs.Per-workflow convenience tools kept (additive — no behavior removed).
glama.jsonat repo root (maintainerorieg) for Glama server verification.Docs (
mcp.md) document the tool model; new tests cover the resolve/validate/execute helpers.Verification
black/isort/mypyclean.run_workflow {who: Alice}→hi Alice; per-workflow tool →hi Bob.Follow-up (needs maintainer's Glama login — Part 2/3 of the playbook)
After this merges + a release: claim the server on glama.ai, sync, configure the build (point
--dirat an empty dir so only the four polished meta-tools are graded), build + publish a release to get the score, then append the Glama score badge to the awesome-mcp-servers PR entry (held pending the badge).