diff --git a/CHANGELOG.md b/CHANGELOG.md
index 1b07554f..f09f2651 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,6 +2,98 @@
 
 All notable changes to OpenAnt are documented in this file.
 
+## [2026-05-07] — Incremental scans + scan pipeline rewire
+
+### Changed
+
+- **`openant scan` rewired end-to-end.** The full pipeline is now
+  explicit and visible in `--help`:
+  `init → parse → app-context → enhance → analyze → verify →
+  build-output → dynamic-test → report`. A single run-mode decision
+  (full vs incremental) is resolved once — at `init` or at `scan` —
+  and recorded in a per-run `meta.json` whose status field
+  transitions `running → success | failed | interrupted`. Step verbs
+  (`parse`, `enhance`, …) read that file and inherit the mode, so a
+  standalone step after `init --incremental` filters correctly
+  without re-passing flags. Docker is preflighted before any work
+  begins, so a missing daemon is reported up front instead of after
+  a full pipeline run.
+- **Dynamic testing is on by default.** The opt-in `--dynamic-test`
+  flag has been replaced by an opt-out `--skip-dynamic-test`. `scan`
+  now produces dynamic verification of findings out of the box;
+  callers without Docker should pass `--skip-dynamic-test`.
+
+### Added
+
+- **Incremental scans.** New `openant diff` subcommand and
+  `--diff-base`, `--pr`, and `--diff-scope` flags on `scan` and
+  `parse` scope the pipeline to changed files only. Go computes a
+  `diff_manifest.json` from the working tree (or a fetched PR head)
+  and threads it through every stage — parse → enhance → analyze →
+  report — so each phase only processes what changed. The diff range
+  surfaces in the CSV export, `_summary.json`, the standalone HTML
+  report, the report header, and the live scan banner.
+- **Explicit run-mode flags on `init`, `scan`, and `parse`** —
+  `--full`, `--incremental`, `--diff-base <ref>`, `--pr <n>`, and
+  `--diff-scope`. With a baseline present, an interactive TTY shows
+  a recap prompt (default Enter = full, the safer choice); non-TTY
+  callers without a flag get a loud error so CI scripts fail fast
+  instead of silently picking a mode.
+
+### Fixed
+
+- **Python call graph no longer drops `self.X()` calls.** The call-
+  graph builder fed function bodies straight into `ast.parse`, but
+  method bodies are stored with their original class indentation —
+  so `ast.parse` raised `IndentationError`, the regex fallback ran
+  without `self.X()` resolution, and every method-to-method call in
+  every Python codebase silently disappeared from the graph. On
+  dbt-core that meant 2,243 of 3,116 functions (72%) marked isolated
+  and a real scan returning zero findings. The fix dedents a
+  temporary copy before parsing and leaves the stored source byte-
+  correct so the disclosure renderer's offsets stay valid.
+- **Disclosure code is byte-faithful to source.** The disclosure
+  renderer pulls the actual file slice from the repo instead of
+  rerunning an LLM rewrite, so every finding's `Vulnerable Code`
+  block matches the real source.
+- **No more silent 401s.** `openant set-api-key` validates the key
+  on save and fails loudly on bad input. `openant scan` prints a
+  blocking warning and exits non-zero when zero API calls succeed,
+  so an all-401 run can no longer masquerade as a clean repo.
+- **CWE tagging is systematic.** `pipeline_output.json` carries
+  non-null `cwe`, `cwe_id`, and `vulnerability_type` for every
+  finding. The Stage 1 prompt asks for them directly rather than
+  relying on the renderer LLM to infer them from prose.
+- **Repo metadata reaches every report envelope.** Repo name,
+  commit SHA, and file count are threaded into `parse.report.json`
+  and `scan.report.json` instead of being lost between stages,
+  eliminating the `[NOT PROVIDED]` placeholders.
+- **`Verified` column reflects the highest evidence tier.**
+  `dynamic` > `verified` > `static`, so dynamically reproduced
+  findings show as `dynamic` and the disclosure footer reads
+  "Confirmed via dynamic test" where applicable.
+- **Call-graph-aware deduplication.** When two findings share a
+  sink/vector and the call graph records an edge between them,
+  they collapse into a single finding.
+- **Dedup matches on CWE** instead of `attack_vector` text, so
+  small wording differences no longer split what's logically the
+  same finding.
+- **Dynamic test Docker context is complete on the first try.**
+  `openant dynamic-test` pre-stages the vulnerable source file
+  into the Docker build context end-to-end through the dynamic-
+  test chain — first-try builds no longer fail because the source
+  isn't in context.
+- **Concurrency-safe Docker resources.** Docker image and network
+  names get a UUID prefix so parallel dynamic-test workers can't
+  collide.
+- **Agreement filter checks the final verdict** instead of the
+  intermediate `agree` flag, so high-confidence dynamic results
+  aren't dropped by a stale agreement signal.
+- **Report prompts respect non-interactive runs.** Prompt output
+  goes to stderr (keeping stdout clean for piped JSON) and the
+  prompt is skipped entirely when there's no TTY, so CI/scripted
+  invocations no longer hang.
+
 ## [2026-04-29] — Python parser dedent fix
 
 ### Fixed