docs: add AI_DISCLOSURE.md and link from README

AI_DISCLOSURE.md

@@ -0,0 +1,103 @@
+---
+tools-used:
+  - OpenCode
+models-used:
+  - claude-opus-4-8
+  - claude-sonnet-4-6
+  - claude-sonnet-5
+  - gpt-5.6-sol
+providers:
+  - Anthropic
+  - OpenAI
+scope:
+  human-authored: >
+    Repository policy and governance text, and final human review of
+    all merged changes.
+  ai-generated: >
+    Shell wrapper scripts (bin/, lib/), the `azoidcapp` Python helper,
+    documentation (docs/), and CI/lint configuration — all with human
+    review and validation.
+last-updated: 2026-07-18
+---
+
+# AI Disclosure
+
+This file describes how AI tools are used in this repository.
+
+## Disclosure levels
+
+This repository uses the vocabulary defined by the
+[W3C AI Content Disclosure Community Group](https://www.w3.org/community/ai-content-disclosure/)
+(a community-group effort, not a ratified W3C standard).
+Four levels are defined:
+
+| Level | Meaning |
+| --- | --- |
+| `none` | No AI tools were used. |
+| `ai-assisted` | AI contributed, but a human authored and reviewed. |
+| `ai-generated` | AI generated the content; a human reviewed it. |
+| `autonomous` | AI produced the content without human review. |
+
+Per the community group's "absence = unknown" principle, files
+without an explicit disclosure tag do not imply a specific level.
+
+## Scope of AI use in this repository
+
+**Repository policy and governance text** — including this disclosure
+file — is human-authored. The final text is authored and validated by
+human contributors.
+
+**Source code in this repository is largely AI-generated:** AI tools
+produce the initial implementation of the sandboxed AI CLI wrapper
+scripts (`bin/bwopencode`, `bin/bwcodex`, `bin/bwclaude`,
+`bin/bwcopilot`) and the shared `lib/bwrap_sandbox_lib.sh` sandbox
+library, as well as `bin/gh-protect-branch`, `bin/azoidcapp`,
+`bin/pemdecompose`, `lib/gpg-passwd.sh`, associated documentation, and
+CI/lint configuration. A human contributor reviews, tests, and
+validates all content before merging; the human remains accountable
+for every merged change.
+
+## Tools and models
+
+AI assistance in this repository is provided via
+**OpenCode** using Anthropic models for implementation and OpenAI GPT-5.6-sol
+for critical code review. See the metadata block at the top of this file for
+the current list of models.
+
+Per-commit attribution uses the `Assisted-by: AGENT:MODEL` trailer
+format. Those per-commit trailers are the authoritative record of
+which model contributed to a specific change.
+
+## Purpose of use
+
+AI tools are used to:
+
+- Draft and refactor the bubblewrap-sandboxed AI CLI wrapper scripts
+  (`bwopencode`, `bwcodex`, `bwclaude`, `bwcopilot`) and the shared
+  `bwrap_sandbox_lib.sh` library.
+- Draft and refactor `gh-protect-branch`, `azoidcapp`, `pemdecompose`,
+  and `gpg-passwd.sh`.
+- Perform critical code review (GPT-5.6-sol).
+- Generate commit messages, code review responses, and documentation.
+
+## Input data / datasets
+
+Only repository source files, public documentation, and
+non-sensitive prompts are provided as input to AI models.
+
+## Limitations and known biases
+
+- LLM output may contain errors, hallucinations, or bias
+  inherited from training data.
+- All AI-generated content is independently reviewed and validated
+  by a human contributor before it is merged.
+- The models listed above do not have access to internal BNL
+  systems, live data, or non-public information unless explicitly
+  provided in a prompt.
+
+## Reviewer disclaimer
+
+AI-generated content in this repository has been reviewed.
+Inclusion of AI-generated material does not substitute for human
+accountability: the contributing staff member is responsible for the
+correctness and appropriateness of every merged change.

README.md

@@ -194,3 +194,8 @@ Requires only `openssl`.
 
 See the per-script doc pages under [`docs/`](docs/) for full usage,
 options, and caveats.
+
+## AI disclosure
+
+This repository's contents are largely AI-generated with human
+review. See [`AI_DISCLOSURE.md`](AI_DISCLOSURE.md) for details.