Land the completed Edison sweep: 353/353 reports (#183)

.github/workflows/pytest.yaml

@@ -8,6 +8,9 @@ on:
       - "tests/**"
       - "pyproject.toml"
       - ".github/workflows/pytest.yaml"
+      # test_run_trait_graph_audit.py::test_the_tracked_corpus_is_clean reads
+      # research/traits/ directly, so an artifact-only PR has to run it (#250).
+      - "research/**"
   push:
     branches: [main]
     paths:
@@ -16,6 +19,9 @@ on:
       - "tests/**"
       - "pyproject.toml"
       - ".github/workflows/pytest.yaml"
+      # test_run_trait_graph_audit.py::test_the_tracked_corpus_is_clean reads
+      # research/traits/ directly, so an artifact-only PR has to run it (#250).
+      - "research/**"
   workflow_dispatch:
 
 jobs:

.github/workflows/qc.yaml

@@ -34,6 +34,13 @@ on:
       # version string in every page footer.
       - "data/embeddings/**"
       - "data/raw/metpo.owl"
+      # research/ joined the gate in #242, via audit-research-artifacts: it
+      # checks every manifest `ok` row's artifact is present and that no
+      # artifact carries a malformed CURIE. Without this line the job cannot
+      # fire on a PR that adds only research artifacts — which is the shape of
+      # every sweep batch, and the shape the gate exists for. Same
+      # filter-narrower-than-the-job bug as #184 and #200 (#250).
+      - "research/**"
   push:
     branches: [main]
   workflow_dispatch:

justfile

@@ -638,10 +638,19 @@ audit-derived-reports:
     fi
     echo "=== derived reports: all current ==="
 
+# Integrity gate for the tracked sweep artifacts: every manifest `ok` row's
+# report is on disk, and no artifact carries a malformed CURIE. Deliberately
+# credential-free and network-free (see run_trait_graph_audit.py --verify), so it
+# runs on a fresh clone and in CI — the two places where a lost artifact is
+# actually noticed. Plain `uv run`, no `--extra dev`: --verify makes no calls and
+# so does not need deep-research-client.
+audit-research-artifacts:
+    uv run python scripts/run_trait_graph_audit.py --verify
+
 # Composite QC: strict closed-schema validation + schema-quality probes +
 # writers audit + proposal citation bar. Mirrors the qc target in
 # MediaIngredientMech / CultureMech.
-qc: pr-sanity validate-strict audit-schema audit-writers audit-proposals audit-graphs audit-justfile-paths audit-derived-reports
+qc: pr-sanity validate-strict audit-schema audit-writers audit-proposals audit-graphs audit-justfile-paths audit-derived-reports audit-research-artifacts
 
 # --- id↔label correspondence gate (vendored byte-identical across the Mech repos) ---
 

research/traits/ecology/commensalism-deep-research-falcon.md.citations.md

@@ -0,0 +1,103 @@
+# Citations for Research Query
+
+**Query:** # Microbial Trait Causal Graph Research Template
+
+## Target Trait
+- **Trait label:** commensalism
+- **Trait identifier (already a CURIE — quote it verbatim, do not prefix it):** traitmech:000042
+- **Trait category:** ECOLOGY
+- **Term kind:** CLASS
+- **Mapping status:** REVIEWED
+- **Definition:** A symbiosis in which the microorganism benefits from the association (e.g. resources, shelter, transport) while the host's fitness remains essentially unaffected.
+- **Parent traits:** traitmech:000040
+- **Synonyms:** commensal
+- **Existing evidence:** DOI:10.1038/s41579-021-00550-7:  (Drew et al. place commensalism on the parasite-mutualist continuum as a near-neutral host interaction.) | DOI:10.1073/pnas.1218525110:  (McFall-Ngai et al. support commensal colonization as a major class of host-associated microbial lifestyles.)
+- **Existing causal graph summary:** commensalism_neutral_host: 11 nodes, 10 edges
+
+## Research Objective
+
+Research the microbial trait **commensalism** as a candidate TraitMech causal graph.
+Focus on source-backed mechanistic entities and causal edges that can be curated into
+`data/traits/ecology/commensalism.yaml`.
+
+## Required Findings
+
+### 1. Trait Scope
+- Clarify what phenotype, physiological capacity, environmental preference, or assay-observed
+  property the trait represents.
+- Identify boundary cases and distinguish the trait from nearby traits.
+
+### 2. Causal Graph Entities
+- Pathways and metabolic modules.
+- Environmental factors and experimental factors.
+- Genes, proteins, enzymes, transporters, and complexes.
+- Chemicals, electron donors, electron acceptors, nutrients, metabolites, and inhibitors.
+- Organelles, cellular localizations, molecular functions, and biological processes.
+
+### 3. Evidence-Backed Edges
+- Propose causal edges as subject-predicate-object triples.
+- For every proposed edge, provide a reference, a short supporting quote/snippet, and notes
+  explaining how the source supports the edge.
+- Prefer DOI references. Use PMID only when a DOI is not available.
+- Mark weak, taxon-specific, assay-specific, or inferred claims as uncertain.
+
+### 4. Ontology Grounding
+- Suggest CURIEs where available: METPO, GO, CHEBI, ENVO, NCBITaxon, EC, UniProt, Rhea,
+  KEGG, MetaCyc, or other stable identifiers.
+- Do not invent identifiers. Label-only candidate nodes are acceptable when grounding is unclear.
+
+## Output Format
+
+Return a curation-focused report with:
+- A short scope summary.
+- Candidate nodes grouped by type.
+- Candidate causal edges in a table with reference, snippet, and notes.
+- DOI-first bibliography.
+- Warnings for claims that should not yet be curated into TraitMech.
+
+**Provider:** falcon
+**Generated:** 2026-08-04T14:55:16.876254
+
+1. drew2021microbialevolutionand pages 1-2
+2. muramatsu2024nutrientacquisitionstrategies pages 2-4
+3. chen2024themicrobiotaa pages 3-5
+4. doranga2024nutritionofescherichia pages 2-4
+5. muramatsu2024nutrientacquisitionstrategies pages 6-7
+6. froismartins2024candidaalbicansvirulence pages 2-4
+7. furuichi2024commensalconsortiadecolonize pages 1-2
+8. torres2024sheddinglighton pages 9-10
+9. froismartins2024candidaalbicansvirulence pages 1-2
+10. furuichi2024commensalconsortiadecolonize pages 3-4
+11. doranga2024nutritionofescherichia pages 6-8
+12. mcfallngai2013animalsina pages 3-4
+13. torres2024sheddinglighton pages 3-5
+14. wilde2024hostcontrolof pages 15-17
+15. wilde2024hostcontrolof pages 21-24
+16. drew2021microbialevolutionand pages 3-4
+17. 10.1128/ecosalplus.esp-0006-2023
+18. 10.1016/j.chom.2024.05.011
+19. 10.1126/science.adi3338
+20. 10.1007/s40588-024-00235-8
+21. 10.1038/s41586-024-07960-6
+22. 10.1128/mbio.00390-24
+23. 10.3389/fmicb.2024.1417864
+24. 10.1038/s41579-021-00550-7
+25. 10.1073/pnas.1218525110
+26. https://doi.org/10.1128/ecosalplus.esp-0006-2023
+27. https://doi.org/10.1016/j.chom.2024.05.011
+28. https://doi.org/10.1126/science.adi3338
+29. https://doi.org/10.1007/s40588-024-00235-8
+30. https://doi.org/10.1038/s41586-024-07960-6
+31. https://doi.org/10.1128/mbio.00390-24
+32. https://doi.org/10.3389/fmicb.2024.1417864
+33. https://doi.org/10.1038/s41579-021-00550-7
+34. https://doi.org/10.1073/pnas.1218525110
+35. https://doi.org/10.1038/s41579-021-00550-7,
+36. https://doi.org/10.1007/s40588-024-00235-8,
+37. https://doi.org/10.1016/j.chom.2024.05.011,
+38. https://doi.org/10.3389/fmicb.2024.1417864,
+39. https://doi.org/10.1128/ecosalplus.esp-0006-2023,
+40. https://doi.org/10.1126/science.adi3338,
+41. https://doi.org/10.1038/s41586-024-07960-6,
+42. https://doi.org/10.1128/mbio.00390-24,
+43. https://doi.org/10.1073/pnas.1218525110,