Add AI-generated detailed model explanations, metadata, and audit for many published BNGL models

[Psuedonerd]

Motivation

• Provide machine-generated, human-readable documentation and machine-actionable metadata_aigenerated.yaml for a large set of published BNGL models to improve discoverability and curation.
• Capture model structure, parameters, molecule/site/state inventories, rule inventories, observables, and workflows in per-model *_aigenerated_detailed.md and metadata_aigenerated.yaml artifacts.
• Produce a consolidated feature audit (data/yaml_feature_audit_4.md) and update skills descriptions to reflect the new explanatory tooling and naming conventions.

Description

• Added many AI-generated detailed model explanations under Published/*/*_aigenerated_detailed.md and shorter explanations *_aigenerated.md, plus corresponding metadata_aigenerated.yaml files for numerous published models, and removed a few legacy coder files that were replaced or consolidated.
• Introduced a batch audit data/yaml_feature_audit_4.md summarizing detected YAML/BNGL feature corrections and created new or updated skill documents under skills/ (renamed explain-rulehub-model-coders to explain-rulehub-model-detailed and metadata_auditing to metadata-auditing).
• Preserved formatting and added compatibility entries in generated YAMLs (e.g., uses_generate_network, uses_multiple_identical_sites, default_sim_command) based on BNGL grammar analysis and inserted concise citation metadata when available.

Testing

• Performed automated YAML syntax validation across the generated metadata_aigenerated.yaml files and basic Markdown render smoke checks for the new .md files; both validations completed successfully.
• Ran a repository-wide static audit that produced data/yaml_feature_audit_4.md documenting detection results and proposed metadata insertions/changes.
• No full CI simulations (network generation or numerical simulation runs) were executed as part of this documentation-focused change.

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

RuleHub/Published/BaruaFceRI2012/metadata_aigenerated.yaml

Line 3 in 779ab2c

description: "FcεRI signaling"

Add citations to every generated metadata file

This is one of 23 new metadata_aigenerated.yaml files with no citation mapping, although the updated audit workflow requires every generated YAML to contain a four-digit year, a reference, and either a PMID or fallback URL. These files therefore remain incomplete as audited publication metadata and provide downstream users no verified paper identity; add the verified citation fields or the documented fallback and report any unresolved match.

---

RuleHub/Published/Barua2013/metadata_aigenerated.yaml

Line 45 in 779ab2c

default_sim_command: simulate_ode

Normalize simulation commands to schema values

For consumers validating these files against metadata-schema.yaml, simulate_ode is not an accepted default_sim_command: the schema permits method tokens such as ode, ssa, and nf, while this batch writes action names such as simulate_ode and simulate in most generated files. Normalize the detected action to the schema's method token, or update the schema and its consumers in the same change, so the audited copies do not fail metadata validation.

---

RuleHub/Published/Chattaraj2021/metadata_aigenerated.yaml

Line 22 in 779ab2c

source:

Emit null when no simulation command exists

When the BNGL has no active simulation command, the audited field is omitted rather than written as null; this occurs here and in ChylekTCR2014, Goldstein1980, Massole2023, and Mertins2023. The updated workflow defines this field as string | null and requires every audited field to be present, so omission makes an inspected no-command result indistinguishable from an unaudited file.

---

RuleHub/Published/Barua2013/metadata_aigenerated.yaml

Line 1 in 779ab2c

id: "Barua_2013"

Derive output names from collaborator metadata

In directories where collaborator-curated metadata takes precedence, such as this directory's Barua_2013_metadata.yaml, the generated copy is still named metadata_aigenerated.yaml. This loses which of the two source YAMLs was audited and can collide with an audit derived from the standard metadata.yaml; use Barua_2013_metadata_aigenerated.yaml and apply the same source-derived naming to the other affected collaborator-metadata directories.

---

RuleHub/data/yaml_feature_audit_4.md

Line 8 in 779ab2c

- Review change: `default_sim_command: ode` → `simulate_ode` — active `simulate_ode(...)` action.

Document citation evidence in the audit report

Even for Barua2013, whose generated YAML already contains a PMID, this per-model section ends after compatibility changes and provides none of the required citation year, PMID/reference provenance, or direct PubMed link; the same omission affects all 30 sections. Consequently the report cannot substantiate the audited citations, and its 27-item review total also excludes unresolved citation work; add the required citation bullets and recalculate the summary totals.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".