Consolidate overlapping Nextflow skills (41 → 28)

.github/workflows/validate.yml

@@ -29,6 +29,9 @@ jobs:
       - name: Install Claude Code
         run: npm install -g "@anthropic-ai/claude-code@${CLAUDE_CODE_VERSION}"
 
+      - name: Test skill consolidation
+        run: python3 -I tools/test_consolidate_skills.py
+
       - name: Check provenance hashes and manifest versions
         run: python3 -I tools/check_sources.py
 

README.md

@@ -3,7 +3,7 @@
 Build, migrate, test, debug and run [Nextflow](https://www.nextflow.io) pipelines with
 Claude, connected to [Nextflow Platform](https://seqera.io), Wave containers and nf-core.
 
-The plugin bundles 41 skills and the hosted Seqera MCP server (`https://mcp.seqera.io/mcp`).
+The plugin bundles 28 skills and the hosted Seqera MCP server (`https://mcp.seqera.io/mcp`).
 
 ## Install
 
@@ -43,11 +43,11 @@ To make the plugin available to everyone working in a repository, add it to that
 
 | Area | Skills |
 | --- | --- |
-| Build pipelines | `create-workflow`, `build-nextflow-pipeline`, `nf-pipeline-design`, `nextflow-config`, `nextflow-schema`, `nextflow-output-patterns`, `search-existing-modules`, `run-module`, `create-container` |
-| Convert to Nextflow | `convert-python-script`, `convert-r-script`, `convert-jupyter-notebook`, `migrate-from-snakemake`, `audit-conversion-readiness`, `triage-pipeline-parameters`, `find-alternative-tools` |
-| Migrate and modernise | `migrate-nextflow-code`, `nextflow-26-syntax`, `nf-migrate-25-04`, `nf-schema-migration`, `nf-v2-boolean-params`, `nf-plugin-development`, `nf-plugin-legacy-migration`, `maintain-nf-core-pipeline` |
-| Test and debug | `nf-test`, `repair-nf-test`, `repair-workflow`, `nf-debug`, `debug-local-run`, `debug-seqera-failed-run`, `nextflow-history`, `nf-run-history`, `nf-data-lineage`, `nf-docker-scripts` |
-| Nextflow Platform | `launch-workflow`, `seqera-mcp`, `ce-credentials-setup`, `seqera-data-links`, `seqerakit` |
+| Build pipelines | `build-nextflow-pipeline`, `nf-pipeline-design`, `nextflow-config`, `nextflow-schema`, `nextflow-output-patterns`, `search-existing-modules`, `run-module`, `create-container` |
+| Convert to Nextflow | `build-nextflow-pipeline` (Python, R and notebook playbooks), `migrate-from-snakemake`, `audit-conversion-readiness`, `nextflow-schema` (parameter triage), `find-alternative-tools` |
+| Migrate and modernise | `migrate-nextflow-code` (version, strict-syntax and boolean references), `nextflow-schema` (schema migration), `nf-plugin-development`, `nf-plugin-legacy-migration`, `maintain-nf-core-pipeline` |
+| Test and debug | `nf-test` (authoring and failure repair), `repair-workflow` (including static diagnostics), `debug-local-run`, `debug-seqera-failed-run`, `nextflow-history` (history, cache and narrative recaps), `nf-data-lineage`, `nf-docker-scripts` |
+| Nextflow Platform | `launch-workflow` (with shared MCP connection reference), `ce-credentials-setup`, `seqera-data-links`, `seqerakit` |
 | General | `nextflow-development`, `install-nextflow` |
 
 ## Licensing
@@ -65,6 +65,7 @@ The plugin content is generated from a host-neutral Agent Plugins package. To up
 ```bash
 python3 -I tools/import_package.py /path/to/nextflow.zip
 python3 -I tools/check_sources.py
+python3 -I tools/test_consolidate_skills.py
 claude plugin validate --strict .
 ```
 
@@ -73,7 +74,15 @@ regenerates `.claude-plugin/plugin.json`, `.claude-plugin/marketplace.json` and
 Do not edit those paths by hand; change the upstream package instead.
 
 The importer also drops the skills listed in `EXCLUDED_SKILLS` and applies the text edits in
-`EDITS`, so the remaining skills don't reference removed ones.
+`EDITS`. It then replays `tools/skill_consolidations.json`: removed entrypoints become linked
+playbooks under their surviving skill, including supporting files and original provenance.
+Change that plan (and any referenced curation assets), rather than editing generated skills.
+To apply a new consolidation to an already imported tree, run:
+
+```bash
+python3 -I tools/consolidate_skills.py
+python3 -I tools/check_sources.py
+```
 
 To release, bump `VERSION` in `tools/import_package.py`, re-import, merge, then tag:
 

skills/audit-conversion-readiness/SKILL.md

@@ -22,7 +22,7 @@ pipeline, input data and execution environment from their request or current
 context. Use user-selected locations for files and results. Consult companion
 skills, bundled references and optional helpers by name when available.
 
-For Seqera operations, use the `seqera-mcp` skill. Any `platform_*` examples
+For Seqera operations, read the [MCP connection](../launch-workflow/references/seqera-mcp/README.md) connection reference. Any `platform_*` examples
 describe API operations. Discover their exact names and parameter schemas with
 `search_seqera_api`, then invoke `call_seqera_api` using those schemas. The host
 manages OAuth for the connected MCP server.
@@ -145,16 +145,16 @@ exists to prevent.
 
 With the audit answered, route onward:
 
-- `triage-pipeline-parameters` — turn the surviving paths and
+- [parameter triage](../nextflow-schema/references/triage-pipeline-parameters/README.md) — turn the surviving paths and
   constants into a parameter surface and schema
 - `create-container` or `nf-docker-scripts` — tools
   with no usable image
 - `search-existing-modules` — tools that may already have a module
 - `find-alternative-tools` — tools blocked by license, GPU
   requirements, or abandonment
 - `seqera-data-links` — cloud data the pipeline must read
-- Then the conversion skill itself: `convert-python-script`, `convert-r-script`,
-  `convert-jupyter-notebook`, `migrate-from-snakemake`, or `create-workflow`
+- Then the conversion skill itself: [Python conversion](../build-nextflow-pipeline/references/convert-python-script/README.md), [R conversion](../build-nextflow-pipeline/references/convert-r-script/README.md),
+  [notebook conversion](../build-nextflow-pipeline/references/convert-jupyter-notebook/README.md), `migrate-from-snakemake`, or [registry composition](../build-nextflow-pipeline/references/create-workflow/README.md)
 
 ## Guardrails
 
@@ -165,5 +165,5 @@ With the audit answered, route onward:
 - Never treat a missing file as a reason to drop the step from the conversion —
   it is a question for the user.
 - Hardcoded paths are findings, not bugs to fix in place. The pipeline's
-  parameter surface is where they get resolved, in `triage-pipeline-parameters`.
+  parameter surface is where they get resolved, in [parameter triage](../nextflow-schema/references/triage-pipeline-parameters/README.md).
 - Keep the audit read-only. No edits to the source material, no scaffolding.