Conciseness

The body is dense and tool-focused with no filler — each entry is parameters, defaults, and gotchas (e.g., "GTEx requires versioned GENCODE IDs", "first hit may not match"), and the deprecated eQTL Catalogue note with its reason is properly quarantined. It sits at 4 rather than 5 because portions restate knowledge Claude already has: the histone-mark dictionary ("H3K4me3 = active promoter... H3K9me3 = constitutive heterochromatin") and bivalent promoter explanation are standard epigenomics, and ReMap_get_transcription_factor_binding and jaspar_search_matrices are each documented twice (Reasoning sections plus Phases 2/6) — minor over-explanation that could be trimmed.

Actionability

Fully executable guidance throughout: seven runnable Python snippets with real arguments and expected result shapes (e.g., `tu.tools.GTEx_get_single_tissue_eqtls(gencode_id=["ENSG00000012048"])` with a Counter over `result["data"]`), exact parameter names with REQUIRED/default annotations, concrete ID formats ("ENSG00000012048.20", "chr17_43705621_T_C_b38"), and result-structure comments. This matches the 5 anchor 'copy-paste ready code; specific examples cover the common cases'.

Workflow Clarity

Clear sequencing with explicit feedback loops: 'Classify the Question First' routes four question types to tool sets, Phases 0–8 give the execution order, and error recovery is explicit — "if 0 results, retry with assay_type='polyA plus RNA-seq'", "if 0 results, append 'tissue', 'epithelium', or 'cell'", plus a Fallback Strategies table and a synthesis rule requiring 3+ converging layers before high confidence. These retry/verification loops match the 5 anchor's 'explicit validation steps; feedback loops for error recovery'; the destructive/batch cap does not apply since all operations are read-only lookups.

Progressive Disclosure

No bundle files exist (references/, scripts/, assets/ are all absent), so this scores on the single file's internal organization — well-sectioned headers, a question-type router up top, phases, and a fallback table. Good structure with minor gaps: tool entries are brief enough to justify staying inline, but there is redundancy from describing tools in both the Reasoning sections and the Phase sections (jaspar, ReMap, UCSC cCREs each appear twice), which a split into a reference file could resolve — fitting the 4 anchor 'most content is appropriately placed; minor organization gaps' rather than 5's cleanly split overview.

Specificity

The description lists several concrete analysis capabilities — "Histone-modification ChIP-seq, ATAC-seq accessibility, chromatin state, and TF binding analysis from ENCODE, Roadmap Epigenomics, ChIP-Atlas" — plus concrete query modes like "chromatin-state-by-tissue queries, TF-binding-by-region, regulatory landscape mapping". It falls just short of the 5 anchor because major capabilities the body actually covers (GTEx eQTL analysis, RegulomeDB variant scoring) are absent, a minor gap in coverage rather than the 'lists several specific actions' (4) vs 'comprehensive coverage' (5) boundary — the omission of eQTL/variant scoring keeps it at 4.

Completeness

Both questions are explicitly answered with concrete trigger phrases: the 'what' is "Histone-modification ChIP-seq, ATAC-seq accessibility, chromatin state, and TF binding analysis from ENCODE..." and the 'when' is an explicit "Use for chromatin-state-by-tissue queries, TF-binding-by-region, regulatory landscape mapping, and ENCODE-cCRE annotations" clause, directly matching the 5 anchor's pattern. Not 4, because the 'when' is fully explicit and enumerates concrete query types rather than being merely present.

Trigger Term Quality

Good keyword coverage of natural domain terms users would say: "ChIP-seq", "ATAC-seq", "chromatin state", "TF binding", "ENCODE", "cCRE", "DNA methylation", "RNA-seq". Not a 5 because common natural variations are missing — "histone mark", "enhancer", "promoter", "eQTL", "regulatory variant" never appear in the description text, so coverage is good but not comprehensive (matching the 4 anchor 'a few natural terms missing').

Distinctiveness Conflict Risk

Clear niche (chromatin accessibility / TF binding / regulatory elements) with explicit boundary disambiguation: "For DNA methylation use tooluniverse-epigenomics; for RNA-seq use tooluniverse-rnaseq-deseq2" actively routes adjacent requests away. Distinct data-type triggers minimize overlap risk, matching the 5 anchor 'clear niche with distinct triggers; minimal conflict risk'.