diff --git a/awesome-ai-auto-research.html b/awesome-ai-auto-research.html
index be6efe5..ef66643 100644
--- a/awesome-ai-auto-research.html
+++ b/awesome-ai-auto-research.html
@@ -793,7 +793,7 @@
Paper Collection
S1
Idea Generation
- 40 methods
+ 45 methods
Expand
@@ -809,7 +809,9 @@ Paper Collection
Evaluation |
Links |
-
+
+
+
| 1 |
PosterMELD |
arXiv'26 |
@@ -829,8 +831,7 @@ Paper Collection
-
-
+
| LLM Internal Knowledge-Based Generation |
@@ -1258,6 +1259,56 @@ Paper Collection
+
+ | 41 |
+ HypoForge |
+ arXiv'26 |
+ Multi-Agent |
+ Reusable scientific skills across generation and testing |
+
+
+ |
+
+
+ | 42 |
+ SGHA |
+ arXiv'26 |
+ Agent |
+ Corpus-first structural-gap discovery on a local LLM |
+
+
+ |
+
+
+ | 43 |
+ MAIL |
+ arXiv'26 |
+ Agent |
+ Memory-driven incremental literature-grounded hypotheses |
+
+
+ |
+
+
+ | 44 |
+ RATIO |
+ arXiv'26 |
+ Benchmark |
+ Relevance defined by three typed ideation moves |
+
+
+ |
+
+
+ | 45 |
+ Lit2Test |
+ arXiv'26 |
+ Benchmark |
+ Each proposal precommits its own falsifying outcome |
+
+
+ |
+
@@ -1271,7 +1322,7 @@ Paper Collection
S2
Literature Review
- 58 methods
+ 65 methods
Expand
@@ -1901,6 +1952,76 @@ Paper Collection
+
+ | 59 |
+ DAS |
+ arXiv'26 |
+ Agent |
+ Stateful closed loop for publication-oriented surveys |
+
+
+ |
+
+
+ | 60 |
+ Crase |
+ arXiv'26 |
+ Agent |
+ Bounded graph exploration; 3x recall@50 at 1/3 cost |
+
+
+ |
+
+
+ | 61 |
+ ITER |
+ arXiv'26 |
+ Retrieval |
+ Retriever trained on agent interaction trajectories |
+
+
+ |
+
+
+ | 62 |
+ DeepWeaver |
+ arXiv'26 |
+ Agent |
+ Weaves fragmented evidence into well-cited answers |
+
+
+ |
+
+
+ | 63 |
+ AgentR |
+ arXiv'26 |
+ Agent |
+ Durable, recoverable literature-review artifacts |
+
+
+ |
+
+
+ | 64 |
+ Tree-of-Concerns |
+ EMNLP'26 |
+ Multi-Agent |
+ Skeptic personas surface unstated limitations |
+
+
+ |
+
+
+ | 65 |
+ Multi-Aspect Retrieval |
+ arXiv'26 |
+ Benchmark |
+ Same-paper retrieval queried from different aspects |
+
+
+ |
+
@@ -1912,7 +2033,7 @@ Paper Collection
S3
Coding & Experiments
- 45 methods
+ 51 methods
Expand
@@ -2420,6 +2541,66 @@ Paper Collection
+
+ | 46 |
+ ReproAgent |
+ EMNLP'26 |
+ Agent |
+ Contract-guided recovery of the split specification |
+
+
+ |
+
+
+ | 47 |
+ DeepRepro |
+ CIKM'26 |
+ Agent |
+ State-aware subplanning over evolving repositories |
+
+
+ |
+
+
+ | 48 |
+ Experimental Fidelity |
+ arXiv'26 |
+ Evaluation |
+ Documents silent methodological hallucinations |
+
+
+ |
+
+
+ | 49 |
+ Praxist |
+ arXiv'26 |
+ Agent |
+ Solution lineages trace gains to causes |
+
+
+ |
+
+
+ | 50 |
+ Skill-Based Baselines |
+ MICCAI'26 |
+ Agent |
+ Literature-guided competitive baselines |
+
+
+ |
+
+
+ | 51 |
+ SciCoQA |
+ ACL'26 |
+ Evaluation |
+ Quality assurance for paper-code alignment |
+
+
+ |
+
@@ -3215,7 +3396,7 @@ Paper Collection
S6
Peer Review
- 48 methods
+ 50 methods
Expand
@@ -3735,6 +3916,26 @@ Paper Collection
+
+ | 49 |
+ Metag |
+ arXiv'26 |
+ Dataset |
+ 349 action items tied to review-rebuttal paper diffs |
+
+
+ |
+
+
+ | 50 |
+ VERA-RL |
+ EMNLP'26 |
+ Training |
+ RL for proactive error detection; VERA-13K |
+
+
+ |
+