Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions .github/workflows/deploy-docs.yml
Original file line number Diff line number Diff line change
Expand Up @@ -6,9 +6,9 @@ on:
- main
paths:
- "website/**"
- "assert/architecture.svg"
- "assert/benchmark_overview.svg"
- "assert/case_study.svg"
- "assets/architecture.svg"
- "assets/benchmark_overview.svg"
- "assets/case_study.svg"
- "scripts/build-pages-site.sh"
- ".github/workflows/deploy-docs.yml"
workflow_dispatch:
Expand Down
16 changes: 8 additions & 8 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -12,14 +12,14 @@
[![Website](https://img.shields.io/badge/website-live-5B5BD6.svg?style=flat-square)](https://bigai-nlco.github.io/belief-context-graph/)
[![Documentation](https://img.shields.io/badge/documentation-online-0F766E.svg?style=flat-square&logo=readthedocs&logoColor=white)](https://belief-context-graph.docs.buildwithfern.com/)
<br>
[![WeChat](https://img.shields.io/badge/WeChat-Community-07C160.svg?style=flat-square&logo=wechat&logoColor=white)](assert/bcg_qr_code.png)
[![Discord](https://img.shields.io/badge/Discord-Community-5865F2.svg?style=flat-square&logo=discord&logoColor=white)](assert/bcg_discard_qr_code.png)
[![WeChat](https://img.shields.io/badge/WeChat-Community-07C160.svg?style=flat-square&logo=wechat&logoColor=white)](assets/bcg_qr_code.png)
[![Discord](https://img.shields.io/badge/Discord-Community-5865F2.svg?style=flat-square&logo=discord&logoColor=white)](assets/bcg_discard_qr_code.png)

</div>

<div align="center">
<a href="assert/benchmark_overview.svg">
<img src="assert/benchmark_overview.svg" width="100%" alt="Full-dataset benchmark comparison with Accuracy grouped by benchmark on the left, mean Token Cost grouped by benchmark on the right, and Default-versus-BCG token-horizon curves below">
<a href="assets/benchmark_overview.svg">
<img src="assets/benchmark_overview.svg" width="100%" alt="Full-dataset benchmark comparison with Accuracy grouped by benchmark on the left, mean Token Cost grouped by benchmark on the right, and Default-versus-BCG token-horizon curves below">
</a>
</div>
<div align="left"><sub><sub>Agent: gpt-5.6-luna with low-effort thinking · Graph/Summary: gpt-5.6-luna with thinking disabled · Harness: <a href="agent-cli/packages/agent-core/src/harness">repository harness</a></sub></sub></div>
Expand Down Expand Up @@ -50,7 +50,7 @@ Belief Context Graph (`BCG`) upgrades agent memory from **retrieval memory** to
- **Temporal Awareness:** Run-based lifecycle with sessions and timestamps — know when each belief was formed and how it evolved
- **Relation Linking:** Forward and backward relationship edges between beliefs, forming a casual decision graph/trace.



## Live Demo

Expand Down Expand Up @@ -100,8 +100,8 @@ The first run walks you through model credentials, context mode, and Graph Const
BCG is an optional context layer between an Agent and its model. In BCG mode, the initial user input and recent completed turns remain in the raw context, while older completed turns stream into Graph Construction; the resulting belief snapshot is then injected into the system prompt. Both the HTTP service and the Python SDK use the same backend registry, construction pipeline, confidence semantics, and graph artifacts.

<p align="center">
<a href="assert/architecture.svg">
<img src="assert/architecture.svg" width="100%" alt="BCG architecture: Agent context management and Python SDK feed unified or hybrid Graph Construction backends, which share an incremental belief graph pipeline and return Graph snapshots to the Agent system prompt">
<a href="assets/architecture.svg">
<img src="assets/architecture.svg" width="100%" alt="BCG architecture: Agent context management and Python SDK feed unified or hybrid Graph Construction backends, which share an incremental belief graph pipeline and return Graph snapshots to the Agent system prompt">
</a>
</p>

Expand All @@ -113,7 +113,7 @@ BCG is an optional context layer between an Agent and its model. In BCG mode, th
This successful BrowseComp case (`browsecomp-0836`) shows Kimi K3 using belief identities and confidence from BCG instead of repeating an already completed search.

<p align="center">
<img src="assert/case_study.svg" alt="BrowseComp task, injected graph beliefs, and selected Kimi K3 thinking passages with graph references highlighted in red" width="100%">
<img src="assets/case_study.svg" alt="BrowseComp task, injected graph beliefs, and selected Kimi K3 thinking passages with graph references highlighted in red" width="100%">
</p>

---
Expand Down
File renamed without changes
File renamed without changes
File renamed without changes
File renamed without changes
File renamed without changes
File renamed without changes
Binary file added assets/case_study.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
File renamed without changes
File renamed without changes
File renamed without changes
6 changes: 3 additions & 3 deletions scripts/build-pages-site.sh
Original file line number Diff line number Diff line change
Expand Up @@ -14,9 +14,9 @@ cp "$ROOT_DIR/website/favicon.svg" "$OUTPUT_DIR/favicon.svg"
cp "$ROOT_DIR/website/site.webmanifest" "$OUTPUT_DIR/site.webmanifest"
cp "$ROOT_DIR/website/index.html" "$OUTPUT_DIR/404.html"

cp "$ROOT_DIR/assert/benchmark_overview.svg" "$OUTPUT_DIR/assets/benchmark-overview.svg"
cp "$ROOT_DIR/assert/architecture.svg" "$OUTPUT_DIR/assets/architecture.svg"
cp "$ROOT_DIR/assert/case_study.svg" "$OUTPUT_DIR/assets/case-study.svg"
cp "$ROOT_DIR/assets/benchmark_overview.svg" "$OUTPUT_DIR/assets/benchmark-overview.svg"
cp "$ROOT_DIR/assets/architecture.svg" "$OUTPUT_DIR/assets/architecture.svg"
cp "$ROOT_DIR/assets/case_study.svg" "$OUTPUT_DIR/assets/case-study.svg"

: > "$OUTPUT_DIR/.nojekyll"

Expand Down
6 changes: 3 additions & 3 deletions scripts/plot_benchmark_overview.py
Original file line number Diff line number Diff line change
Expand Up @@ -67,11 +67,11 @@ def compose(summary: Path, horizon: Path, output: Path) -> None:
def main() -> None:
parser = argparse.ArgumentParser()
parser.add_argument(
"--summary", type=Path, default=Path("assert/benchmark_summary.svg")
"--summary", type=Path, default=Path("assets/benchmark_summary.svg")
)
parser.add_argument("--horizon", type=Path, default=Path("assert/token_cost.svg"))
parser.add_argument("--horizon", type=Path, default=Path("assets/token_cost.svg"))
parser.add_argument(
"--output", type=Path, default=Path("assert/benchmark_overview.svg")
"--output", type=Path, default=Path("assets/benchmark_overview.svg")
)
args = parser.parse_args()
compose(args.summary, args.horizon, args.output)
Expand Down
2 changes: 1 addition & 1 deletion scripts/plot_benchmark_summary.py
Original file line number Diff line number Diff line change
Expand Up @@ -280,7 +280,7 @@ def render_svg(output: Path) -> None:
def main() -> None:
parser = argparse.ArgumentParser()
parser.add_argument(
"--output", type=Path, default=Path("assert/benchmark_summary.svg")
"--output", type=Path, default=Path("assets/benchmark_summary.svg")
)
args = parser.parse_args()
render_svg(args.output)
Expand Down
2 changes: 1 addition & 1 deletion scripts/plot_token_horizon.py
Original file line number Diff line number Diff line change
Expand Up @@ -375,7 +375,7 @@ def parse_args() -> argparse.Namespace:
parser = argparse.ArgumentParser()
parser.add_argument("--browsecomp-results", type=Path, required=True)
parser.add_argument("--browsecomp-zh-results", type=Path, required=True)
parser.add_argument("--output", type=Path, default=Path("assert/token_cost.svg"))
parser.add_argument("--output", type=Path, default=Path("assets/token_cost.svg"))
parser.add_argument("--horizon", type=int, default=22)
return parser.parse_args()

Expand Down
Loading