From c2a581ddb59d09055dbef7dd15120bfdd60303ef Mon Sep 17 00:00:00 2001
From: Theseus <theseus@agents.livingip.xyz>
Date: Tue, 10 Mar 2026 16:16:12 +0000
Subject: [PATCH] theseus: add full-text request list for 12 active inference
 papers

- 7 open access (arXiv/MDPI/Frontiers) -- need re-fetch for full text
- 5 paywalled -- Cory can grab via Google Scholar

Pentagon-Agent: Theseus <25B96405-E50F-45ED-9C92-D8046DFAAD00>
---
 inbox/full-text-requests.md | 60 +++++++++++++++++++++++++++++++++++++
 1 file changed, 60 insertions(+)
 create mode 100644 inbox/full-text-requests.md

diff --git a/inbox/full-text-requests.md b/inbox/full-text-requests.md
new file mode 100644
index 0000000..f5f1ce5
--- /dev/null
+++ b/inbox/full-text-requests.md
@@ -0,0 +1,60 @@
+# Full Text Requests
+
+Papers archived during Opus research sessions where only abstracts/reconstructions were captured. Full PDFs would significantly improve extraction quality.
+
+## Priority: HIGH (core to active inference operationalization)
+
+1. **Friston et al. 2024 -- Designing Ecosystems of Intelligence from First Principles**
+   - URL: https://journals.sagepub.com/doi/10.1177/26339137231222481
+   - arXiv: https://arxiv.org/abs/2212.01354
+   - Status: Open access via arXiv
+   - Why: THE bullseye paper. Multi-agent active inference for AI ecosystems.
+
+2. **Friston et al. 2024 -- Federated Inference and Belief Sharing**
+   - URL: https://www.sciencedirect.com/science/article/pii/S0149763423004694
+   - Status: Likely paywalled (ScienceDirect)
+   - Why: Formalizes belief-sharing between agents. Validates our PR review as federated inference.
+
+3. **Kaufmann et al. 2021 -- An Active Inference Model of Collective Intelligence**
+   - URL: https://www.mdpi.com/1099-4300/23/7/830
+   - Status: Open access (MDPI Entropy)
+   - Why: Shows CI emerges from simple agent capabilities (ToM + goal alignment).
+
+4. **Ruiz-Serra et al. 2024 -- Factorised Active Inference for Strategic Multi-Agent Interactions**
+   - URL: https://arxiv.org/abs/2411.07362
+   - Status: Open access (arXiv)
+   - Why: Critical corrective -- individual optimization != collective optimization.
+
+5. **Kagan et al. 2025 -- As One and Many: Group-Level Active Inference**
+   - URL: https://www.mdpi.com/1099-4300/27/2/143
+   - Status: Open access (MDPI Entropy)
+   - Why: Formal conditions for group-level agency via Markov blankets.
+
+## Priority: MEDIUM
+
+6. **Vasil et al. 2020 -- A World Unto Itself: Human Communication as Active Inference**
+   - URL: https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2020.00417/full
+   - Status: Open access (Frontiers)
+   - Why: Communication as joint active inference. Grounds chat-as-sensor.
+
+7. **Sajid et al. 2021 -- Active Inference: Demystified and Compared**
+   - URL: https://direct.mit.edu/neco/article/33/3/674/97486
+   - Status: Likely paywalled (MIT Press)
+   - Why: Best comparison of active inference vs RL for agent design.
+
+8. **Albarracin et al. 2024 -- Shared Protentions in Multi-Agent Active Inference**
+   - URL: https://www.mdpi.com/1099-4300/26/4/303
+   - Status: Open access (MDPI Entropy)
+   - Why: Shared expectations between agents.
+
+## Priority: LOW (supporting)
+
+9. **Friston 2015 -- Active Inference and Epistemic Value** -- paywalled (PubMed 25689102)
+10. **Ramstead et al. 2018 -- Answering Schrodingers Question** -- paywalled (Physics of Life Reviews)
+11. **Da Costa et al. 2020 -- Active Inference on Discrete State-Spaces** -- paywalled (ScienceDirect)
+12. **Orchestrator 2025 -- Active Inference for Multi-Agent LLM Systems** -- open access (arXiv 2509.05651)
+
+## Notes
+- 7 of 12 are open access -- may just need WebFetch on full HTML/PDF
+- 5 paywalled -- Cory can grab via Google Scholar
+- Once full text added to archive files, re-run extraction for richer claims