Matt Silverlock

This commit **updates documentation** for **`bonk` fix requests** within the **`opencode` agents** system. It specifically modifies `.opencode/agents/bonk.md` to reflect an "implementation-first" approach for these requests. This **maintenance** task ensures that the documentation accurately describes the current design and development strategy, improving clarity for users and developers interacting with this feature.

This commit performs a **maintenance update** to the **Bonk workflow configurations**, involving significant renaming and model changes. The workflow previously known as `bonk.yml` has been renamed to `bigbonk.yml` and now defaults to the `claude-opus-4-6` AI model, alongside an updated trigger command. Concurrently, the workflow previously known as `lilbonk.yml` has been renamed to `bonk.yml` and now defaults to the `kimi-k2.5` AI model, with modifications to exclude `/bigbonk` mentions. These updates **redefine the default AI models** and operational parameters for the respective Bonk automation tasks, impacting their execution and output characteristics.

This commit **introduces new capabilities** by adding the **Kimi K2.5 model configuration** to the `cloudflare-ai-gateway` provider, detailing its specific capabilities and limits. Concurrently, it establishes a **new GitHub Actions workflow** named `lilbonk.yml` to automate tasks based on issue and pull request comments. This work enhances both the **available AI models** for integration and the **project's CI/CD automation**, streamlining development and expanding AI service offerings.

This commit introduces a **refactoring** to the **`bonk.yml` GitHub Actions workflow** to **improve its early exit condition**. Specifically, it optimizes the workflow's execution by ensuring it **exits more efficiently** when a "mention" (e.g., a specific keyword or user tag) is not present. This change prevents unnecessary subsequent steps from running, thereby **reducing CI/CD resource consumption** and speeding up feedback cycles for irrelevant triggers. The **`bonk` workflow** will now be more performant and resource-conscious.

This commit refines the **CI/CD workflow** by adjusting the trigger conditions for the `bigbonk` job. Specifically, the `.github/workflows/bigbonk.yml` file was updated to ensure the `bonk` job only executes if the triggering comment explicitly contains the `/bigbonk` command. This **maintenance** change improves the precision of workflow invocations, **preventing unintended runs** and making the automated process more controlled and efficient. It impacts how developers interact with the `bigbonk` automation, requiring a specific command for execution.

This commit performs a **maintenance chore** by updating the `fetch-depth` configuration within the **GitHub Actions workflow** responsible for `new-pr-review`, located at `.github/workflows/new-pr-review.yml`. The `fetch-depth` is increased to 30, meaning the workflow will now fetch the last 30 commits of the repository history. This adjustment provides more historical context for automated review tools and scripts, ensuring they have sufficient data to perform accurate comparisons and analyses during pull request checks.

This commit **improves** the **PR review prompt** within the **GitHub Actions workflow** defined in `.github/workflows/new-pr-review.yml`. Specifically, it **adds the pull request number and repository information** to the prompt's context. This **enhancement** provides better steering for the automated review process, aiming to generate more relevant and accurate feedback by supplying crucial contextual details.

This commit **enhances the `bonk` automated PR reviewer** by updating its prompt to include a new rule. The modification to `.github/bonk_reviewer.md` instructs the reviewer to explicitly confirm it is acting on the correct issue or pull request. This **feature improvement** aims to **increase the accuracy and contextual relevance of automated review comments**, preventing misdirected feedback. Consequently, it will lead to a more reliable and efficient **code review workflow** by ensuring the bot's comments are always on-topic.

Introduces a new GitHub Actions workflow (`bigbonk.yml`) that provides a new operational capability for handling larger tasks, triggerable via `/bigbonk`.

This is an isolated, minor update to the `bonk.yml` configuration file, with no production impact.

This commit updates permissions for the new PR review GitHub workflow, granting `write` access to resolve a functional issue. This ensures the workflow can now execute its tasks correctly.

Affects the internal review prompt used by the "bonk reviewer" process. This is an isolated documentation cleanup, removing an explicit reference with no production impact.

This is an isolated refactoring of internal GitHub workflows, specifically how PR review prompts are managed. It has no impact on the product or external users.

This commit updates the GitHub Actions workflow for new PR reviews, making automated prompts suggestion-only and adding a timeout. It impacts the development process by refining automated feedback and improving workflow reliability, with no direct production impact.

This commit introduces new GitHub Actions workflows, adding automated PR review capabilities triggered by comments or automatically, thereby enhancing the development workflow with new CI/CD automation.

This commit **enhances the Changelog page** by **limiting the number of rendered entries** on `src/pages/changelog/index.astro` to a maximum of 50. This **performance optimization** aims to improve page load times and reduce visual clutter for users browsing the changelog. To maintain full accessibility to historical data, an **empty state message** is now displayed when no entries are found, providing a direct link to search for older or specific changelog items. This change significantly improves the initial user experience of the **Changelog feature** while guiding users to more comprehensive search options.

This commit **refactors** the **`review-code-examples` command** by replacing its direct agent reference with a more structured **`code-review` skill**. The configuration in `.opencode/command/review-code-examples.md` was updated to specify a new model and include clearer instructions for code editing within documentation reviews. This change primarily impacts the **AI agent configuration** and **LLM interaction** for documentation reviews. The sharpened command wording aims to improve the **LLM's understanding and execution** of code review tasks, leading to more precise and effective feedback for users leveraging this AI-powered feature.

This commit **refactors** the **`changelog` command** by **migrating its core logic** to a newly introduced **agent skill**. The existing command, defined in `.opencode/command/changelog.md`, now delegates its operations to this new skill, which is detailed in `.agents/skills/changelog/SKILL.md`. This **new capability** centralizes the management of product changelog entries, encompassing creation, updates, and reviews, within the agent skill framework. The change **simplifies the command's internal implementation** and enhances the modularity and extensibility of changelog-related functionalities.

Fixes a write permission issue within the `bonk` module's content handling. This is an isolated fix.

This commit primarily **fixes** an issue causing the **`bonk` GitHub Actions job** to get stuck indefinitely in CI by implementing a 30-minute timeout. Complementing this, the **agent's `code-review` skill documentation** is updated to direct the agent to use local `node_modules` or existing search tools for references, reducing reliance on external fetches that could contribute to CI hangs. This **maintenance** work also introduces a new `opencode.jsonc` configuration file with default permissions and experimental settings. Together, these changes enhance CI stability and streamline agent operations by promoting local resource utilization.