refactor(standards): split language_standards.py into a per-language package

.claude/agents/class-start-scout.md

@@ -1,13 +1,13 @@
 ---
 name: class-start-scout
-description: Mechanical diff-reporting and verification-sequence execution for epic #1295 (extending gitgalaxy/core/detector.py's _CLASS_START_NAMED_EXTRACTION_LANGS allowlist). Runs tests/tools/class_start_diff.py and the tree_sitter_accuracy_audit.py/crucible_check.py/audit_check.py verification sequence, reports structured results -- it does not judge whether an extra/missing name is a real bug, decide how to fix a regex, or edit the allowlist itself. Use before a per-language triage pass (to get one language's exact extra/missing name diff without reading language_standards.py cold) and after a regex fix + allowlist flip (to run the full verification sequence and report pass/fail).
+description: Mechanical diff-reporting and verification-sequence execution for epic #1295 (extending gitgalaxy/core/detector.py's _CLASS_START_NAMED_EXTRACTION_LANGS allowlist). Runs tests/tools/class_start_diff.py and the tree_sitter_accuracy_audit.py/crucible_check.py/audit_check.py verification sequence, reports structured results -- it does not judge whether an extra/missing name is a real bug, decide how to fix a regex, or edit the allowlist itself. Use before a per-language triage pass (to get one language's exact extra/missing name diff without reading language_standards/languages/<lang>.py cold) and after a regex fix + allowlist flip (to run the full verification sequence and report pass/fail).
 tools: Bash, Read
 model: haiku
 ---
 
 You run scripts and report their output cleanly. You do not judge whether a `class_start` regex
 is buggy, decide whether an extra/missing name is a real bug versus a ground-truth scope mismatch,
-or edit `gitgalaxy/core/detector.py`/`gitgalaxy/standards/language_standards.py` yourself -- that
+or edit `gitgalaxy/core/detector.py`/`gitgalaxy/standards/language_standards/` yourself -- that
 judgment belongs to whoever asked for the scout pass (the main conversation or a case-authoring
 subagent). If something looks ambiguous, report it plainly rather than guessing what it means.
 

.claude/agents/strict-signature-scout.md

@@ -1,6 +1,6 @@
 ---
 name: strict-signature-scout
-description: Mechanical gap-reporting and candidate-verification for epic #1069 (hardening tests/extraction/languages/test_<lang>_strict.py). Runs tests/extraction/tools/audit_strict_coverage.py and tests/extraction/tools/verify_candidates.py and reports structured results -- it does not draft test cases or judge realism, only executes scripts and formats their output. Use before a case-writing pass (to get one language's exact gap list without loading language_standards.py cold) and after drafting candidate cases (to batch-verify them against the real compiled regex before they're written into a test file).
+description: Mechanical gap-reporting and candidate-verification for epic #1069 (hardening tests/extraction/languages/test_<lang>_strict.py). Runs tests/extraction/tools/audit_strict_coverage.py and tests/extraction/tools/verify_candidates.py and reports structured results -- it does not draft test cases or judge realism, only executes scripts and formats their output. Use before a case-writing pass (to get one language's exact gap list without loading language_standards/languages/<lang>.py cold) and after drafting candidate cases (to batch-verify them against the real compiled regex before they're written into a test file).
 tools: Bash, Read
 model: haiku
 ---

.claude/skills/harden-language-extraction/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: harden-language-extraction
-description: Deepen or fix a language's structural-extraction accuracy in gitgalaxy/standards/language_standards.py (func_start/args/class_start/_dependency_capture regexes) using the epic #813 methodology. Use when the user asks to "harden extraction for X", "add strict tests for language X", "find bugs in X's parsing", or similar per-language regex-correctness work -- not for adding a brand-new language from scratch (that's how_to_add_a_language.md's LLM generation prompt) and not for the broader ReDoS/boundary-correctness rules covered there either.
+description: Deepen or fix a language's structural-extraction accuracy in gitgalaxy/standards/language_standards/languages/<lang>.py (func_start/args/class_start/_dependency_capture regexes) using the epic #813 methodology. Use when the user asks to "harden extraction for X", "add strict tests for language X", "find bugs in X's parsing", or similar per-language regex-correctness work -- not for adding a brand-new language from scratch (that's how_to_add_a_language.md's LLM generation prompt) and not for the broader ReDoS/boundary-correctness rules covered there either.
 ---
 
 Source of truth is `tests/extraction/how_to_harden_extraction.md` -- read it directly, don't work
@@ -12,7 +12,8 @@ a language-specific quirk.
 
 ## Process (translating that doc's 5-stage pipeline to this tool's Agent tool)
 
-1. **Load the target.** Read the language's full `rules` dict from `language_standards.py` and its
+1. **Load the target.** Read the language's full `rules` dict from
+   `gitgalaxy/standards/language_standards/languages/<lang>.py` and its
    existing cases in `tests/extraction/languages/test_<lang>.py` (or the old monolithic
    `test_*_extraction_strict.py` files if it hasn't migrated yet). Work one language, all four
    gauntlets, in one sitting -- not gauntlet-by-gauntlet across languages.
@@ -30,7 +31,7 @@ a language-specific quirk.
    unrealistic payload), not an automatic fix.
 5. **Fix real bugs with the full discipline**: ReDoS scaling check on any quantifier change,
    `python tests/tools/audit_check.py` (add `--regenerate` for pure line-shifts), then
-   `pytest tests/extraction/languages/test_<lang>.py`. If the fix touches `language_standards.py`,
+   `pytest tests/extraction/languages/test_<lang>.py`. If the fix touches `language_standards/`,
    `detector.py`, or `prism.py`: `python tests/tools/crucible_check.py` -- confirm any real diff is
    confined to the language(s) actually changed (check `_dependency_capture` fixes for legitimate
    cross-language DAG ripple, per the doc's note) before `--update --yes`.

.github/workflows/tree-sitter-accuracy-audit.yml

@@ -18,7 +18,7 @@ on:
     paths:
       - "gitgalaxy/core/detector.py"
       - "gitgalaxy/core/prism.py"
-      - "gitgalaxy/standards/language_standards.py"
+      - "gitgalaxy/standards/language_standards/**"
       - "tests/tools/tree_sitter_accuracy_audit.py"
       - "tests/tree_sitter_accuracy_baseline_*.json"
       - ".github/workflows/tree-sitter-accuracy-audit.yml"
@@ -63,8 +63,8 @@ jobs:
         run: |
           cd gitgalaxy
           python tests/tools/tree_sitter_accuracy_audit.py --summary-table
-          if ! git diff --exit-code -- gitgalaxy/standards/language_standards.py; then
-            echo "::error::gitgalaxy/standards/language_standards.py's summary table is stale."
+          if ! git diff --exit-code -- gitgalaxy/standards/language_standards/__init__.py; then
+            echo "::error::gitgalaxy/standards/language_standards/__init__.py's summary table is stale."
             echo "Run 'python tests/tools/tree_sitter_accuracy_audit.py --summary-table' and commit the result."
             exit 1
           fi

.github/workflows/tree-sitter-accuracy-history.yml

@@ -10,7 +10,7 @@ name: Tree-sitter Accuracy History
 # baselined language again and append one row each to
 # docs/self_scan/tree_sitter_accuracy_history.csv (an accumulating time series, unlike the
 # gitignored self-scan DB -- this is meant to be graphed later), regenerate the Markdown
-# summary table embedded in language_standards.py's docstring so it can't drift from the
+# summary table embedded in language_standards/__init__.py's docstring so it can't drift from the
 # committed baselines between now and the next manual --summary-table run, and re-render
 # docs/self_scan/tree_sitter_accuracy_chart.svg (a small-multiples bar chart of the most
 # recent history batch -- always the one this same job just appended). Opens an auto-merged
@@ -19,7 +19,7 @@ name: Tree-sitter Accuracy History
 # protection), and why the loop guard below is load-bearing.
 #
 # Adaptive by design, not just by trigger path: the trigger below fires on any push touching
-# detector.py/prism.py/language_standards.py, but not every such push actually moves a
+# detector.py/prism.py/language_standards/, but not every such push actually moves a
 # language's measured accuracy. --history itself detects that (see
 # tree_sitter_accuracy_audit.py's `_batch_matches_measured`) and skips appending a row when the
 # fresh measurement is identical to the last recorded batch -- which means --chart re-renders
@@ -34,7 +34,7 @@ on:
     paths:
       - "gitgalaxy/core/detector.py"
       - "gitgalaxy/core/prism.py"
-      - "gitgalaxy/standards/language_standards.py"
+      - "gitgalaxy/standards/language_standards/**"
       - "tests/tools/tree_sitter_accuracy_audit.py"
       - "tests/tree_sitter_accuracy_baseline_*.json"
 
@@ -44,7 +44,7 @@ permissions:
 jobs:
   tree-sitter-accuracy-history:
     # Guards against the same infinite-loop shape documented in gitgalaxy.yml's
-    # full-report job: this workflow's own bot commit touches language_standards.py and
+    # full-report job: this workflow's own bot commit touches language_standards/__init__.py and
     # the baseline-adjacent history CSV, which are both in the trigger's `paths` filter,
     # so without this it would re-trigger itself on every merge of its own PR.
     if: |
@@ -114,7 +114,7 @@ jobs:
             ${{ steps.blurbs.outputs.blurbs }}
 
             - Appends one row per language to `docs/self_scan/tree_sitter_accuracy_history.csv`.
-            - Regenerates the summary table in `gitgalaxy/standards/language_standards.py`.
+            - Regenerates the summary table in `gitgalaxy/standards/language_standards/__init__.py`.
             - Re-renders `docs/self_scan/tree_sitter_accuracy_chart.svg` from that same batch.
 
             Never touches `tests/tree_sitter_accuracy_baseline_*.json` -- those stay
@@ -124,7 +124,7 @@ jobs:
           add-paths: |
             docs/self_scan/tree_sitter_accuracy_history.csv
             docs/self_scan/tree_sitter_accuracy_chart.svg
-            gitgalaxy/standards/language_standards.py
+            gitgalaxy/standards/language_standards/__init__.py
 
       # Same rationale as gitgalaxy.yml's "Merge the LLM Brief PR" step -- nothing else
       # gates this PR, so without an explicit merge it just sits mergeable-but-unmerged.

.github/workflows/tri-comparison-audit.yml

@@ -24,7 +24,7 @@ on:
     paths:
       - "gitgalaxy/core/detector.py"
       - "gitgalaxy/core/prism.py"
-      - "gitgalaxy/standards/language_standards.py"
+      - "gitgalaxy/standards/language_standards/**"
       - "tests/tools/tri_comparison_chart.py"
       - "tests/tools/tri_comparison_gatherer.py"
       - "tests/tools/tri_comparison_reconcile.py"

.github/workflows/tri-comparison-history.yml

@@ -16,7 +16,7 @@ name: Tri-Comparison History
 # whatever actually changed -- same mechanism gitgalaxy.yml uses for the LLM architecture brief.
 #
 # Adaptive by design, not just by trigger path: the trigger below fires on any push touching
-# detector.py/prism.py/language_standards.py, but not every such push moves a language's measured
+# detector.py/prism.py/language_standards/, but not every such push moves a language's measured
 # numbers or an existing ledger entry's counts. Unlike the tree-sitter history CSV (an
 # append-only file that needs its own _batch_matches_measured dedup check to stay a no-op), the
 # three files here are overwritten in place -- "no diff" already IS the adaptivity, so
@@ -29,7 +29,7 @@ on:
     paths:
       - "gitgalaxy/core/detector.py"
       - "gitgalaxy/core/prism.py"
-      - "gitgalaxy/standards/language_standards.py"
+      - "gitgalaxy/standards/language_standards/**"
       - "tests/tools/tri_comparison_chart.py"
       - "tests/tools/tri_comparison_gatherer.py"
       - "tests/tools/tri_comparison_reconcile.py"

gitgalaxy/standards/README.md

@@ -45,7 +45,7 @@ Each file in this directory serves a distinct calibration purpose for the downst
 
 * **`gitgalaxy_config.py` (Global Ingestion Firewall):** Defines Zero-Trust ingestion boundaries. It houses the Supply Chain Firewall configurations (approved vs. blacklisted imports), global file denylists, X-Ray binary scanner bypasses, and physical file-size clamps.
 * **`language_lens.py` (Identity Classifier):** The Bayesian engine that assigns definitive "Identity Locks" to files. It defines the multi-tiered confidence hierarchy, resolving extension collisions by weighing exact filename matches against Contextual Baselines.
-* **`language_standards.py` (Structural Signature Registry):** The massive, highly optimized structural mapping registry for 50+ languages. It defines exactly how to slice a language into Branch Logic, State Flux, High-Risk Execution, and Object Declarations using ReDoS-proof regular expressions.
+* **`language_standards/` (Structural Signature Registry):** A package with one file per language under `languages/` (59 today), assembled by `__init__.py` into the `LANGUAGE_DEFINITIONS` registry every downstream engine imports. Each per-language file defines exactly how to slice that language into Branch Logic, State Flux, High-Risk Execution, and Object Declarations using ReDoS-proof regular expressions; cross-language shared pieces (`LENS_CONFIG`, `PRISM_CONFIG`, the `GLOBAL_*` debt/AI-SDK detectors, `PROJECT_OVERRIDES`) live in their own sibling modules. See `how_to_add_a_language.md` for the registration mechanics.
 * **`analysis_lens.py` (Mathematical Constants & Threat Policies):** The repository of Threat Policies, Sigmoid Curve tuning, and K-Means clustering medians. It dictates how raw structural signals are mathematically converted into normalized 0-100% risk exposure vectors.
 * **`how_to_add_a_language.md` (Extension Protocol):** Contains the strict prompt engineering protocols required to generate ReDoS-proof language dictionaries using advanced LLMs, bypassing the need for manual parser development.
 

gitgalaxy/standards/gitgalaxy_config.py

@@ -99,7 +99,7 @@
     "yarn.lock",
     "composer.lock",
     "gitgalaxy/core/aperture.py",
-    "gitgalaxy/standards/language_standards.py",
+    "gitgalaxy/standards/language_standards/",
     "gitgalaxy/security/security_lens.py",
     "gitgalaxy/tools/network_auditing/full_api_network_map.py",
     "gitgalaxy/tools/cobol_to_cobol/cobol_schema_forge.py",

gitgalaxy/standards/how_to_add_a_language.md

@@ -14,16 +14,32 @@ To add a new language to the Language Classifier, you will use an advanced LLM (
 ---
 
 ### Step 1: Initialize the LLM Context
-Before asking the LLM to generate the new language signatures, upload the `gitgalaxy/standards/language_standards.py` file to the chat window. Issue this exact command:
-> *"Read this file to understand how the GitGalaxy Structural Signature Analysis Engine uses bounded regex to guarantee ReDoS immunity. Pay close attention to how C++ and Python are mapped to prevent Catastrophic Backtracking."*
+`gitgalaxy/standards/language_standards/` is a package with one file per language (each a
+`DEFINITION: dict[str, Any] = {...}` literal, assembled into the `LANGUAGE_DEFINITIONS` registry
+by `__init__.py`), not a single monolithic file. Before asking the LLM to generate new language
+signatures, upload `gitgalaxy/standards/language_standards/__init__.py` plus one or two
+representative per-language files (e.g. `languages/python.py` and `languages/cpp.py`) to the chat
+window. Issue this exact command:
+> *"Read these files to understand how the GitGalaxy Structural Signature Analysis Engine uses bounded regex to guarantee ReDoS immunity, and how each language's rules dict is registered in `__init__.py`. Pay close attention to how C++ and Python are mapped to prevent Catastrophic Backtracking."*
 
 ### Step 2: Inject the Structural Signature Prompt
 Copy the **Generation Prompt** below and paste it into the LLM. Replace `[TARGET LANGUAGE]` with the exact language you want to map.
 
 ### Step 3: Register the Signatures
-1. Open `gitgalaxy/standards/language_standards.py`.
-2. Locate the `LANGUAGE_DEFINITIONS` registry.
-3. Paste the generated Python dictionary directly into the registry to instantly grant the engine native support for the new language architecture.
+1. Create `gitgalaxy/standards/language_standards/languages/<lang>.py` containing
+   `DEFINITION: dict[str, Any] = {...}` with the generated dictionary as its value (see any
+   existing file in that directory for the exact shape — license header, `import re`, and a
+   `from .._shared_patterns import ...` line only if the rules reference one of the shared
+   `GLOBAL_*` cross-language constants).
+2. In `gitgalaxy/standards/language_standards/__init__.py`, add an
+   `from .languages import <lang> as _<lang>` import (isort-sorted alphabetically among the
+   others) and a `"<lang>": _<lang>.DEFINITION,` line inside the `LANGUAGE_DEFINITIONS` dict, in
+   the same position other entries follow (insertion order is preserved end-to-end, so add new
+   languages at the end unless you have a specific reason not to).
+3. If the language's key isn't a valid Python identifier (spaces, hyphens, etc. — the only
+   existing case is `"objective-c"`, whose module is `languages/objectivec.py`), the file's own
+   module name still needs to be identifier-safe even though the dict key itself doesn't; pick a
+   concatenated or underscored slug and keep the dict key exactly as the LLM/consumers expect it.
 
 <br><br>