docs(ai-policy): ask for the reasoning effort, make the agent optional

AGENT_MD_SNIPPETS.md

@@ -12,7 +12,7 @@ All contributions generated or assisted by this agent must fully comply with:
 
 ### What this agent must always do
 
-- Add an `Assisted-by: AGENT_NAME:MODEL_VERSION` git trailer to every commit containing AI-assisted content.
+- Add an `Assisted-by: MODEL_VERSION (REASONING_EFFORT)` git trailer to every commit containing AI-assisted content, for example `Assisted-by: claude-opus-4-5 (medium)`. Ask the contributor for the reasoning effort if you do not know it, and leave it out only if the tool has no such setting. Leave out your own agent name unless the contributor asks for it, then write `AGENT_NAME:MODEL_VERSION (REASONING_EFFORT)`.
 - Ensure every pull request includes a disclosure of AI tool use in the PR description.
 - Produce focused, scoped pull requests that address exactly one concern. Do not touch unrelated files or introduce incidental refactors.
 - Verify all dependencies against actual package registries before suggesting them. Do not use hallucinated or unverified package names.

AI_POLICY.md

@@ -20,20 +20,23 @@ Every pull request containing AI-assisted code, documentation, or tests must dec
 For full traceability at the commit level, each commit containing AI-assisted content must include an `Assisted-by:` git trailer:
 
 ```
-Assisted-by: AGENT_NAME:MODEL_VERSION
+Assisted-by: MODEL_VERSION (REASONING_EFFORT)
+Assisted-by: AGENT_NAME:MODEL_VERSION (REASONING_EFFORT)
 ```
 
-The agent name and model version identify the AI tool. Basic development tools such as git, compilers, editors, and static analyzers are not listed - these are standard parts of any development workflow regardless of AI involvement.
+The model version identifies the AI tool; for an open-weight model you run yourself, it includes the quantization. Add the reasoning effort whenever the tool lets you choose one, because the same model gives different results depending on it; leave it out only if the tool has no such setting. The agent name, meaning the tool or harness that ran the model, is optional. Existing commits with `AGENT_NAME:MODEL_VERSION` don't need to be rewritten. `Assisted-by:` covers any kind of help. If an LLM generated content, you can write `AI-Assisted-By:` instead to say so explicitly; the AI Policy check recognises both. Basic development tools such as git, compilers, editors, and static analyzers are not listed - these are standard parts of any development workflow regardless of AI involvement.
 
 The PR description disclosure explains how AI was used; the commit trailer ensures that provenance is permanently recorded in version history and available to future contributors, auditors, and tooling.
 
 Examples:
 
 ```
-Assisted-by: Devstral:devstral-small-2507
-Assisted-by: ClaudeCode:claude-sonnet-4-6
-Assisted-by: Qwen:qwen3-coder-32b
+Assisted-by: claude-opus-4-5 (medium)
+Assisted-by: ClaudeCode:claude-opus-4-5 (high)
 Assisted-by: Copilot:gpt-4o
+Assisted-by: Devstral:devstral-small-2507
+Assisted-by: qwen3-coder-30b-a3b-instruct-q4_k_m
+AI-Assisted-By: claude-opus-4-5 (low)
 ```
 
 ### Author Accountability

CONTRIBUTING.md

@@ -29,7 +29,7 @@ Please read the [Developer Manuals][devmanual] to learn how to create your first
 
 Nextcloud allows contributions made with the help of AI tools. You are the author of everything you submit - AI assistance does not change that responsibility.
 
-* **Disclosure:** Declare AI tool use in the PR description and add an `Assisted-by: AGENT_NAME:MODEL_VERSION` git trailer to each affected commit.
+* **Disclosure:** Declare AI tool use in the PR description and add an `Assisted-by: MODEL_VERSION (REASONING_EFFORT)` git trailer to each affected commit, for example `Assisted-by: claude-opus-4-5 (medium)`. Leave out the effort only if the tool has no such setting; the agent name in front (`AGENT_NAME:`) is optional. If an LLM generated content, `AI-Assisted-By:` says so explicitly.
 
 * **Accountability:** You must be able to explain, defend, and modify every line you submit. If a reviewer asks why something works a certain way, "the AI wrote it" is not an answer.