diff --git a/applications/collatz/collatz_k_scripts/collatz_generate_run_report.py b/applications/collatz/collatz_k_scripts/collatz_generate_run_report.py
index cd918d7..9ddf38b 100644
--- a/applications/collatz/collatz_k_scripts/collatz_generate_run_report.py
+++ b/applications/collatz/collatz_k_scripts/collatz_generate_run_report.py
@@ -87,6 +87,21 @@ class Section7Metrics:
     max_modulus_coverage_pct: float
 
 
+@dataclass(frozen=True)
+class D16D17Metrics:
+    d16_candidates: int
+    d16_cover: int
+    d16_maxA_after: int
+    d17_pairs: int
+    noyau_post_d17_size: int
+
+
+@dataclass(frozen=True)
+class Fusion25Metrics:
+    t_to_valid_clauses: dict[int, int]
+    t_to_domain_size: dict[int, int]
+
+
 def parse_extend_metrics_from_logs(pipeline_extend_log: str, paliers_finale_log: str) -> ExtendMetrics:
     """
     Extract metrics needed for the report.
@@ -160,6 +175,73 @@ def parse_section7_metrics(seuil_global_md: str, coverage_md: str) -> Section7Me
     return Section7Metrics(nstar=nstar, max_modulus_power=max_power, max_modulus_coverage_pct=max_cov)
 
 
+def _parse_int_after_colon(line: str) -> int | None:
+    m = re.search(r":\s*([0-9][0-9_]*)\b", line)
+    if not m:
+        return None
+    return int(m.group(1).replace("_", ""))
+
+
+def parse_d16_d17_metrics(d16_impact_md: str, d17_impact_md: str, noyau_post_d17_json: str) -> D16D17Metrics:
+    d16_candidates = None
+    d16_cover = None
+    d16_maxA_after = None
+    for line in d16_impact_md.splitlines():
+        if "candidats D16" in line:
+            d16_candidates = _parse_int_after_colon(line)
+        if "couverture (avec sœurs)" in line:
+            d16_cover = _parse_int_after_colon(line)
+        if "invariant max A16 après" in line:
+            d16_maxA_after = _parse_int_after_colon(line)
+    if d16_candidates is None or d16_cover is None or d16_maxA_after is None:
+        raise ValueError("Cannot parse D16 metrics from impact md")
+
+    d17_pairs = None
+    for line in d17_impact_md.splitlines():
+        if "paires candidates D17" in line:
+            d17_pairs = _parse_int_after_colon(line)
+    if d17_pairs is None:
+        raise ValueError("Cannot parse D17 metrics from impact md")
+
+    data = read_text(Path(noyau_post_d17_json))
+    j = re.search(r'"noyau"\s*:\s*\[', data)
+    if not j:
+        raise ValueError("Cannot locate noyau list in noyau_post_D17.json")
+    # The file is JSON with {"noyau":[...], "palier": 28}. To avoid loading huge data,
+    # count commas in the noyau array region with a bounded approach (file is ~30MB).
+    # We do a lightweight parse by counting numbers.
+    import json as _json
+
+    obj = _json.loads(data)
+    noyau_post_d17_size = len(obj.get("noyau", []))
+
+    return D16D17Metrics(
+        d16_candidates=d16_candidates,
+        d16_cover=d16_cover,
+        d16_maxA_after=d16_maxA_after,
+        d17_pairs=d17_pairs,
+        noyau_post_d17_size=noyau_post_d17_size,
+    )
+
+
+def parse_fusion25_metrics(md_text_by_t: dict[int, str]) -> Fusion25Metrics:
+    t_to_valid: dict[int, int] = {}
+    t_to_domain: dict[int, int] = {}
+    for t, text in md_text_by_t.items():
+        domain = None
+        valid = None
+        for line in text.splitlines():
+            if "Taille du domaine analysé" in line:
+                domain = _parse_int_after_colon(line)
+            if "Clauses valides" in line:
+                valid = _parse_int_after_colon(line)
+        if domain is None or valid is None:
+            raise ValueError(f"Cannot parse fusion metrics from md for t={t}")
+        t_to_domain[t] = domain
+        t_to_valid[t] = valid
+    return Fusion25Metrics(t_to_valid_clauses=t_to_valid, t_to_domain_size=t_to_domain)
+
+
 def parse_running_command_from_paliers_finale_log(paliers_finale_log: str) -> str:
     """
     Extract the last 'Running: ...' command from paliers_finale.log.
@@ -307,6 +389,104 @@ def write_section7_run_report(
     output_path.write_text("\n".join(lines) + "\n", encoding="utf-8")
 
 
+def write_d16_d17_run_report(
+    *,
+    output_path: Path,
+    report_title: str,
+    command: str,
+    git_commit: str,
+    sha_entries: list[Sha256Entry],
+    metrics: D16D17Metrics,
+    out_dir: Path,
+) -> None:
+    lines: list[str] = []
+    lines.append("**Auteur** : Équipe 4NK")
+    lines.append("")
+    lines.append(f"# {report_title}")
+    lines.append("")
+    lines.append("## Contexte")
+    lines.append("")
+    lines.append("- **But du run** : exécuter le pipeline D16/D17 après fusion (sections “pipeline directe”).")
+    lines.append("- **Assertion ciblée** : produire les CSV D16/D17 et le noyau résiduel `noyau_post_D17.json`.")
+    lines.append("")
+    lines.append("## Code et reproductibilité")
+    lines.append("")
+    if git_commit:
+        lines.append(f"- **Commit Git** : `{git_commit}`")
+    if command:
+        lines.append("- **Commande** :")
+        lines.append("")
+        lines.append("```bash")
+        lines.append(command)
+        lines.append("```")
+        lines.append("")
+    lines.append("## Empreintes sha256 (scripts, artefacts)")
+    lines.append("")
+    lines.extend(format_sha256_list(sha_entries))
+    lines.append("")
+    lines.append("## Compteurs et métriques")
+    lines.append("")
+    lines.append(f"- candidats D16 : {metrics.d16_candidates}")
+    lines.append(f"- couverture D16 (avec sœurs) : {metrics.d16_cover}")
+    lines.append(f"- max A16 après : {metrics.d16_maxA_after}")
+    lines.append(f"- paires candidates D17 : {metrics.d17_pairs}")
+    lines.append(f"- noyau post-D17 (résidus) : {metrics.noyau_post_d17_size}")
+    lines.append("")
+    lines.append("## Chemins d’artefacts")
+    lines.append("")
+    lines.append(f"- OUT : `{out_dir}`")
+    lines.append("")
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+    output_path.write_text("\n".join(lines) + "\n", encoding="utf-8")
+
+
+def write_fusion25_run_report(
+    *,
+    output_path: Path,
+    report_title: str,
+    command: str,
+    git_commit: str,
+    sha_entries: list[Sha256Entry],
+    metrics: Fusion25Metrics,
+    out_dir: Path,
+) -> None:
+    lines: list[str] = []
+    lines.append("**Auteur** : Équipe 4NK")
+    lines.append("")
+    lines.append(f"# {report_title}")
+    lines.append("")
+    lines.append("## Contexte")
+    lines.append("")
+    lines.append("- **But du run** : produire des clauses de fusion F(t) au palier \\(2^{25}\\) (t=11,12,14).")
+    lines.append("")
+    lines.append("## Code et reproductibilité")
+    lines.append("")
+    if git_commit:
+        lines.append(f"- **Commit Git** : `{git_commit}`")
+    if command:
+        lines.append("- **Commande** :")
+        lines.append("")
+        lines.append("```bash")
+        lines.append(command)
+        lines.append("```")
+        lines.append("")
+    lines.append("## Empreintes sha256 (scripts, artefacts)")
+    lines.append("")
+    lines.extend(format_sha256_list(sha_entries))
+    lines.append("")
+    lines.append("## Compteurs et métriques")
+    lines.append("")
+    for t in sorted(metrics.t_to_valid_clauses.keys()):
+        lines.append(f"- t={t} : domaine={metrics.t_to_domain_size[t]}, clauses_valides={metrics.t_to_valid_clauses[t]}")
+    lines.append("")
+    lines.append("## Chemins d’artefacts")
+    lines.append("")
+    lines.append(f"- OUT : `{out_dir}`")
+    lines.append("")
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+    output_path.write_text("\n".join(lines) + "\n", encoding="utf-8")
+
+
 def pick_report_date_from_mtime(paths: list[Path]) -> str:
     mtimes = [p.stat().st_mtime for p in paths if p.exists()]
     if not mtimes:
@@ -323,7 +503,7 @@ def main() -> None:
     ap.add_argument(
         "--profile",
         default="extend_finale",
-        choices=["extend_finale", "validation_section7"],
+        choices=["extend_finale", "validation_section7", "pipeline_d16_d17", "fusion_palier2p25"],
         help="Report profile",
     )
     ap.add_argument("--pipeline-extend-log", default=None, help="Path to pipeline_extend.log (defaults to OUT/pipeline_extend.log)")
@@ -423,6 +603,79 @@ def main() -> None:
         print(f"Wrote: {output_path}")
         return
 
+    if args.profile == "pipeline_d16_d17":
+        command = args.command.strip() if args.command.strip() else "cd applications/collatz && ./scripts/02-run-pipeline.sh"
+        d16_impact_path = out_dir / "candidats_D16_apres_fusion_palier2p27_et_impact.md"
+        d17_impact_path = out_dir / "candidats_D17_apres_fusion_palier2p28_et_impact.md"
+        noyau_post_d17_path = out_dir / "noyaux" / "noyau_post_D17.json"
+        metrics = parse_d16_d17_metrics(
+            read_text(d16_impact_path),
+            read_text(d17_impact_path),
+            str(noyau_post_d17_path),
+        )
+        date_str = pick_report_date_from_mtime([d16_impact_path, d17_impact_path, noyau_post_d17_path])
+        sha_paths: list[Path] = [
+            repo_root / "applications" / "collatz" / "scripts" / "02-run-pipeline.sh",
+            repo_root / "applications" / "collatz" / "collatz_k_scripts" / "reproduce_all_audits.py",
+            repo_root / "applications" / "collatz" / "collatz_k_scripts" / "collatz_k_pipeline.py",
+            d16_impact_path,
+            d17_impact_path,
+            out_dir / "candidats_D16_apres_fusion_palier2p27.csv",
+            out_dir / "candidats_D17_apres_fusion_palier2p28.csv",
+            noyau_post_d17_path,
+        ]
+        sha_entries = compute_sha256_entries(sha_paths)
+        output_path = docs_dir / f"collatz_run_report_{date_str}_{args.scope}.md"
+        write_d16_d17_run_report(
+            output_path=output_path,
+            report_title=f"Rapport d’exécution — {args.scope}",
+            command=command,
+            git_commit=commit_hash,
+            sha_entries=sha_entries,
+            metrics=metrics,
+            out_dir=out_dir,
+        )
+        print(f"Wrote: {output_path}")
+        return
+
+    if args.profile == "fusion_palier2p25":
+        command = args.command.strip() if args.command.strip() else "cd applications/collatz && ./scripts/02-run-pipeline.sh"
+        md_by_t = {
+            11: read_text(out_dir / "fusion_t11_palier2p25.md"),
+            12: read_text(out_dir / "fusion_t12_palier2p25.md"),
+            14: read_text(out_dir / "fusion_t14_palier2p25.md"),
+        }
+        metrics = parse_fusion25_metrics(md_by_t)
+        paths_for_date = [
+            out_dir / "fusion_t11_palier2p25.md",
+            out_dir / "fusion_t12_palier2p25.md",
+            out_dir / "fusion_t14_palier2p25.md",
+        ]
+        date_str = pick_report_date_from_mtime(paths_for_date)
+        sha_paths: list[Path] = [
+            repo_root / "applications" / "collatz" / "scripts" / "02-run-pipeline.sh",
+            repo_root / "applications" / "collatz" / "collatz_k_scripts" / "collatz_k_pipeline.py",
+            out_dir / "fusion_t11_palier2p25.md",
+            out_dir / "fusion_t11_palier2p25.csv",
+            out_dir / "fusion_t12_palier2p25.md",
+            out_dir / "fusion_t12_palier2p25.csv",
+            out_dir / "fusion_t14_palier2p25.md",
+            out_dir / "fusion_t14_palier2p25.csv",
+        ]
+        sha_entries = compute_sha256_entries(sha_paths)
+        output_path = docs_dir / f"collatz_run_report_{date_str}_{args.scope}.md"
+        write_fusion25_run_report(
+            output_path=output_path,
+            report_title=f"Rapport d’exécution — {args.scope}",
+            command=command,
+            git_commit=commit_hash,
+            sha_entries=sha_entries,
+            metrics=metrics,
+            out_dir=out_dir,
+        )
+        print(f"Wrote: {output_path}")
+        return
+
     raise ValueError(f"Unknown profile: {args.profile}")
 
 
diff --git a/docs/collatz_run_report_format.md b/docs/collatz_run_report_format.md
index dead285..3bde6cb 100644
--- a/docs/collatz_run_report_format.md
+++ b/docs/collatz_run_report_format.md
@@ -37,6 +37,26 @@ python3 applications/collatz/collatz_k_scripts/collatz_generate_run_report.py \
   --docs-dir docs
 ```
 
+Pour le pipeline D16/D17 après fusion :
+
+```bash
+python3 applications/collatz/collatz_k_scripts/collatz_generate_run_report.py \
+  --profile pipeline_d16_d17 \
+  --scope pipeline_d16_d17 \
+  --out-dir applications/collatz/out \
+  --docs-dir docs
+```
+
+Pour les fusions au palier \(2^{25}\) (t=11,12,14) :
+
+```bash
+python3 applications/collatz/collatz_k_scripts/collatz_generate_run_report.py \
+  --profile fusion_palier2p25 \
+  --scope fusion_palier2p25 \
+  --out-dir applications/collatz/out \
+  --docs-dir docs
+```
+
 ### Contexte
 
 - **But du run** : (énoncé court)
diff --git a/docs/features/collatz_run_report_generator.md b/docs/features/collatz_run_report_generator.md
index 6b798e0..6985e78 100644
--- a/docs/features/collatz_run_report_generator.md
+++ b/docs/features/collatz_run_report_generator.md
@@ -23,6 +23,7 @@ afin d’éviter toute insertion de transcript terminal dans les documents math
 - Ajout d’un format standard : `docs/collatz_run_report_format.md`.
 - Ajout d’un générateur : `applications/collatz/collatz_k_scripts/collatz_generate_run_report.py`.
 - Ajout d’un exemple réel de rapport : `docs/collatz_run_report_2026-03-04_extend_D18_D21_resume_from_D20.md`.
+ - Profils supportés : `extend_finale`, `validation_section7`, `pipeline_d16_d17`, `fusion_palier2p25`.
 
 ## Modalités d’analyse