Signed-off-by: Rahul Krishna <rkrsn@ibm.com>

Signed-off-by: Rahul Krishna <rkrsn@ibm.com>

Signed-off-by: Rahul Krishna <rkrsn@ibm.com>

Signed-off-by: Rahul Krishna <rkrsn@ibm.com>

Signed-off-by: Rahul Krishna <rkrsn@ibm.com>

Signed-off-by: Rahul Krishna <rkrsn@ibm.com>

- codeanalyzer-ts backend page: lead with an Install section (pip install
  codeanalyzer-typescript / brew) that yields the `cants` CLI; rename binary
  references from codeanalyzer-typescript to cants; note the package-vs-command
  naming; correct the -a row (level 2 = call graph, not CodeQL).
- installing.mdx: add a TypeScript Analysis dependency tab covering the same
  pip/brew install and the cants command.

CodeQL was removed from the analyzer. Reframe the level-2 dataflow enrichment
as planned work backed by Jelly (cs-au-dk/jelly), and note that -a 2 currently
produces the same call graph as -a 1. Scrubs the Call graph, schema example,
Performance, and Current maturity sections.

- Rewrite prose across all docs pages into a neutral, professional
  technical-documentation tone; remove anthropomorphizing/agent-guessing
  framing and marketing phrasing.
- Fix Java method declarations to use fully-qualified parameter types
  (e.g. addOption(org.apache.commons.cli.Option),
  parse(org.apache.commons.cli.Options, java.lang.String[])).
- Remove the reachability section from Core concepts and Common tasks
  (kept as an application-level capability in cocoa).
- Replace the broken Mermaid diagram on "What is CLDK?" and expand the
  language-coverage table.
- Merge the "Guides" and "Build with CLDK" sidebar groups.
- Remove the "Build with CLDK" overview and "COCO MCP Toolbox" pages and
  strip all inbound links.
- Replace em-dashes with parentheses/colon/comma and normalize en-dashes.

Signed-off-by: Rahul Krishna <rkrsn@ibm.com>

Add a top-level "At scale" section documenting the emit/poll architecture:
codeanalyzer-{java,python,typescript} project a namespaced property graph into
a shared Neo4j (--emit neo4j), and the CLDK SDK polls it as a read-only Cypher
client via Neo4jConnectionConfig. Covers the shared-graph schema, idempotent
incremental writes, the --emit schema contract, a Kubernetes CronJob/Deployment
topology, and an end-to-end Odoo (Python + JS) multi-lingual walkthrough.

Reconcile the contradictory TypeScript maturity claims to a single voice: TS is
beta — symbol table and call graph work through CLDK.typescript(...), entrypoint
detection is not yet implemented. Update what-is-cldk, the Python-API reference,
the backends overview, and codeanalyzer-ts. Cross-link existing read-only Neo4j
notes to the new section and correct the Neo4j language list to Java/Python/TS.

…uto-update workflow

Regenerate src/content/docs/reference/python-api/ from cldk 1.2.0 (griffe):
python.md picks up the full 1.2.0 Py* schema, and a new typescript.md documents
TypeScriptAnalysis + the TS* schema (TSApplication, call graph, interfaces,
enums, decorators). Add the TypeScript page to scripts/gen_api_docs.py PAGES with
a house-style authored intro (preserved across regen by the injection markers),
and wire "TypeScript analysis" into the sidebar and the reference index.

Add .github/workflows/update-api-docs.yml: on a python-sdk release (repository_
dispatch), manual dispatch, or a weekly cron, it clones the SDK at the release
tag, regenerates the reference with griffe, and opens a PR when the generated
docs change — keeping the source-controlled API reference matched to the SDK
(the deploy workflow only regenerates ephemerally for the published build).

Also includes an expanded codeanalyzer-python backend reference page.

…tures

Adversarial verification against the backend sources surfaced real errors in
the at-scale/Neo4j docs, now fixed:

- HIGH: the full-run orphan prune is app-anchor-scoped only for Python
  (codeanalyzer-python bolt.py). The Java and TypeScript Bolt writers prune any
  unit/module absent from the run regardless of app, so the docs no longer
  promise shared-DB safety for Java/TS — they direct per-app databases instead.
- Odoo example: cants discovers only .ts/.tsx/.mts/.cts (discovery.ts SOURCE_EXTS),
  so it cannot analyze Odoo's OWL JavaScript frontend. Reframed the multi-lingual
  walkthrough as Odoo (Python addons) + a real TypeScript service, with an honest
  note; fixed the get_callers qualified name (rooted at the emit -i dir).
- Flags: targeted-run is --file-name on canpy, --target-files on codeanalyzer/cants.
- Jelly is experimental and opt-in (--call-graph-provider jelly|both), not
  "planned/not implemented" — corrected all three mentions in codeanalyzer-ts.
- codeanalyzer-python: primary CLI is canpy (codeanalyzer is a deprecated alias);
  added the --emit/--app-name/--neo4j-* options to the table.

Added the "all the goodies" coverage the at-scale page was missing: a raw-Cypher
section (app enumeration, full-text code search over j_code_fts/py_code_fts/
code_fts, Java-only CRUD + system-dependency-graph methods), external/phantom
nodes, an actionable schema_version check, the snapshot-vs-live tradeoff, and a
Neo4j Browser tip. Dropped an unused Badge import.

Findings produced by an adversarial multi-agent verification workflow (22
confirmed against source, 1 refuted).

Drop the weekly cron and manual workflow_dispatch triggers; the workflow now runs
solely on a repository_dispatch (event-type: sdk-release) sent by python-sdk's
release workflow after it publishes, carrying the released tag in
client_payload.ref. Resolve the ref from the dispatch payload (with a latest-tag
fallback) and regenerate + PR the reference against that release.

…page

gen_api_docs.py now reads the installed cldk version (importlib.metadata) and
emits a `cldk <version>` PyPI badge next to the source badge plus an explicit
"API reference generated from cldk <version>." line in each generated block.
Regenerated against cldk 1.2.0, so the version a reader is looking at is now
unambiguous and updates automatically on each release-driven regeneration.