docs(contributing): standardize AI disclosures - #660
Conversation
📝 WalkthroughWalkthroughChangesAI disclosure guidance
Estimated code review effort: 1 (Trivial) | ~3 minutes Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Greptile SummaryStandardizes disclosure requirements for AI-assisted GitHub contributions.
Confidence Score: 4/5The documentation-only change is safe to merge, with a non-blocking ambiguity in how the standardized template represents an unavailable model identifier. The new fallback prose covers unavailable model and version identifiers, but the template explicitly provides an unavailable variant only for version, allowing inconsistent model-field formatting. Files Needing Attention: AGENTS.md Important Files Changed
Reviews (1): Last reviewed commit: "docs(contributing): standardize AI discl..." | Re-trigger Greptile |
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
| When AI contributes GitHub content—including a pull request description, review, pull request | ||
| comment, or discussion post—append this disclosure: | ||
|
|
||
| `*AI-assisted — Tool: <tool>; model: <provider>/<model>; version: <version-or-unavailable>.*` |
There was a problem hiding this comment.
The template requires model: <provider>/<model>, while the following instruction requires unavailable when the model is not exposed. This leaves contributors choosing incompatible forms such as model: unavailable or model: unavailable/unavailable, undermining the intended consistent disclosure format.
Context Used: AGENTS.md (source)
Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@AGENTS.md`:
- Around line 125-128: Update the AI disclosure guidance around the
“AI-assisted” template to require exact runtime-exposed values independently for
tool, provider/model, and version. Specify that any unavailable field, including
tool, must use “unavailable”; prohibit leaving placeholders or inferring
identifiers.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Central YAML (base), Organization UI (inherited)
Review profile: CHILL
Plan: Pro Plus
Run ID: 0c796b71-7806-40de-9070-1b3f06258192
📒 Files selected for processing (1)
AGENTS.md
| `*AI-assisted — Tool: <tool>; model: <provider>/<model>; version: <version-or-unavailable>.*` | ||
|
|
||
| Use the exact model and version identifiers exposed by the runtime. Never infer or guess them; use | ||
| `unavailable` when either value is not exposed. |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
Define the fallback for every identifier, including the tool.
The rule only specifies unavailable for missing model/version values, while <tool> has no equivalent fallback. Require each field to use its exact runtime-exposed value independently, or unavailable when that field is not exposed; never leave placeholders or infer values.
Proposed wording
-Use the exact model and version identifiers exposed by the runtime. Never infer or guess them; use
-`unavailable` when either value is not exposed.
+Use the exact tool, model, and version identifiers exposed by the runtime. Never infer or guess any
+field; use `unavailable` for each identifier that is not exposed.Based on learnings, AI disclosures must include exposed tool, model, and version details, using unavailable for missing identifiers.
📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| `*AI-assisted — Tool: <tool>; model: <provider>/<model>; version: <version-or-unavailable>.*` | |
| Use the exact model and version identifiers exposed by the runtime. Never infer or guess them; use | |
| `unavailable` when either value is not exposed. | |
| `*AI-assisted — Tool: <tool>; model: <provider>/<model>; version: <version-or-unavailable>.*` | |
| Use the exact tool, model, and version identifiers exposed by the runtime. Never infer or guess any | |
| field; use `unavailable` for each identifier that is not exposed. |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@AGENTS.md` around lines 125 - 128, Update the AI disclosure guidance around
the “AI-assisted” template to require exact runtime-exposed values independently
for tool, provider/model, and version. Specify that any unavailable field,
including tool, must use “unavailable”; prohibit leaving placeholders or
inferring identifiers.
Source: Learnings
Instruction counts
No instruction-count regression above 1%. Only instruction counts gate. Wall clock is shown for context — on identical hardware it moves 4-20% run to run. Measured by tak — instruction-counted CLI benchmarks, stored in this repository's git notes.
|
1 participant
Summary
unavailablewhen runtime identifiers are not exposed so agents do not guessImpact
AI-assisted GitHub contributions will carry a consistent, more informative disclosure across all contribution surfaces.
Validation
git diff --checkAI-assisted — Tool: OpenAI Codex; model: OpenAI/GPT-5; version: unavailable.
Note
Low Risk
Documentation-only change to contributor guidelines; no runtime, security, or application code is affected.
Overview
Standardizes how AI-assisted GitHub contributions must be labeled in
AGENTS.mdunder GitHub Interactions.The old rule only required a generic “AI-generated” note on PR comments and discussions. It now applies to pull request descriptions, reviews, comments, and discussion posts, and mandates a single footer format:
*AI-assisted — Tool: <tool>; model: <provider>/<model>; version: <version-or-unavailable>.*Agents must use runtime-exposed tool/model/version identifiers and write
unavailablewhen those values are not provided—no guessing.Reviewed by Cursor Bugbot for commit ef33317. Bugbot is set up for automated code reviews on this repo. Configure here.
Summary by CodeRabbit