-
Notifications
You must be signed in to change notification settings - Fork 5
chore: Test editorial agents workflow #1065
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
base: master
Are you sure you want to change the base?
Conversation
Added Topic Structure Agent documentation for auditing compliance with CTRT framework. Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Added voice-tone guidelines for documentation PRs, focusing on consistent voice, active tone, and present tense usage. Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
…nto justine-claude
Add changelog-formatter.skill.md to enable manual formatting and style checking of Seqera changelogs (Platform, Wave, Nextflow, Fusion). The skill applies universal style rules from the Seqera changelog style guide: - Past tense, active voice, sentence case - Product-specific requirements (attribution for Wave/Nextflow) - Quality checks for consistency and formatting Location: .claudeskills/ (standard location for Claude Code skills)
Add automated formatting and style enforcement for Platform Cloud and Enterprise changelogs. Components added: 1. Manual skill (.claudeskills/platform-changelog-formatter.skill.md) - Format individual Platform changelogs on-demand - Applies component-based organization and universal style rules - Platform-only scope (Cloud and Enterprise) 2. Automated PR review agent (.github/.claude/agents/platform-changelog-style.md) - Automatically reviews Platform changelog changes in PRs - Enforces past tense, active voice, Studios capitalization - Requires component organization (Studios, Compute, Pipelines, etc.) - Fails PR if critical style violations found 3. Review configuration update (.github/.claude/agents/review-config.yaml) - Registered platform-changelog-style agent - Triggers on all Cloud/Enterprise changelog changes - Does NOT apply to Wave/Nextflow/Fusion changelogs Scope: ONLY Seqera Platform Cloud and Enterprise changelogs - changelog/seqera-cloud/*.md - changelog/seqera-enterprise/*.md Wave, Nextflow, and Fusion changelogs remain unchanged (keep attribution format).
Removed changelog-formatter.skill.md as it's redundant with the Platform-specific formatter. We only need platform-changelog-formatter.skill.md which handles Cloud and Enterprise changelogs. Wave, Nextflow, and Fusion changelogs will remain in their current format (with attribution) and don't need automated formatting.
…ASK] Added explicit note in platform-changelog-formatter skill that component headers MUST be level 3 headers (###), NOT bold (**Component**). Updated in two places: - Section 4 (component organization): Added CRITICAL note - Section 7 (quality checks): Clarified the check This ensures components are formatted as: ### Studios NOT: **Studios**
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
Signed-off-by: Justine Geffen <justinegeffen@users.noreply.github.com>
- Move platform-changelog-style.md to .claude/agents/ - Remove duplicate .github/.claude/ directory - Align with Claude Code standard architecture Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Delete ./lint script (linter doesn't exist yet) - Simplify CLAUDE.md to focus on agent-based reviews - Remove all references to ../Claude/Linting/claude-md-linter.js - Keep terminology standards (RNA-Seq rules, product names) - Document /review command usage instead - Agents already handle terminology, capitalization, and structure checks Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Delete .claude/agents/topic-structure.md - Delete .claude/agents/inclusive-language.md - Update review-config.yaml to remove agent configs and related severity items - Update review.md to remove agents from comprehensive and new-content profiles - Update editorial-review/SKILL.md to remove from agent lists - Remove topic-structure-review and inclusive-review jobs from docs-review.yml - Update workflow summary to reflect remaining 4 agents - Remove agents from CLAUDE.md agent list Keeping core agents: voice-tone, terminology, clarity, punctuation, page-structure Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Delete .claude/agents/platform-changelog-style.md - Agent was not connected to any workflows or configs - Functionality is covered by platform-changelog-formatter skill - Skill does the actual formatting work, agent was redundant Active agents now: voice-tone, terminology, clarity, punctuation, page-structure, docs-fix Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Clarify fix commands: mention both /fix-docs and @claude fix for different workflows - Remove absolute path reference (/Users/justine.geffen/work/docs/) - Remove misplaced 'Review output format' section (74 lines) - Section contained agent instructions, not user documentation - Included typo 'Googl Guide' and 'Non-american' - Was redundant with agent descriptions - File now focused on user-facing documentation workflow Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Update docs-review.yml to watch Platform Enterprise and Cloud docs - Update claude-pr-review.yml to watch Platform docs only - Update review-config.yaml agent patterns to match Platform docs Now watches: - platform-enterprise_docs/**/*.md(x) - platform-cloud/docs/**/*.md(x) - platform-enterprise_versioned_docs/**/*.md(x) No longer watches: - fusion_docs, wave_docs, multiqc_docs, platform-api-docs, changelog This focuses automated reviews on Platform documentation only. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Clarify which directories get automated PR reviews - Separate directories into two categories: 1. Automated PR reviews (Platform docs only) 2. Manual review only (other products) - Add file counts for context - Matches actual workflow paths in docs-review.yml and claude-pr-review.yml Now accurately documents that only Platform Enterprise and Cloud docs trigger automated agent reviews, while other products need manual /review. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
❌ Deploy Preview for seqera-docs failed. Why did it fail? →
|
📝 Documentation Review Summary
Review powered by Claude Code SME agents To apply fixes: Comment |
The claude-code-action requires OIDC token to authenticate. Added id-token: write permission to both workflows: - docs-review.yml - claude-pr-review.yml This fixes the error: 'Could not fetch an OIDC token' Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
📝 Claude Editorial ReviewReview completed - check logs for details. Powered by Claude Code editorial-review skill |
The claude-code-action expects 'direct_prompt' not 'prompt'. Updated all workflow steps to use the correct parameter name. This fixes the warning: 'Unexpected input(s) prompt' Changed in both workflows: - docs-review.yml (5 occurrences) - claude-pr-review.yml (2 occurrences) Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
📝 Documentation Review Summary
Review powered by Claude Code SME agents To apply fixes: Comment |
📝 Claude Editorial ReviewReview completed - check logs for details. Powered by Claude Code editorial-review skill |
The claude-code-action posts comments directly to PRs. The Post Structure Results step tried to read 'structure-results.md' which doesn't exist, causing: ENOENT: no such file or directory Removed the redundant post-result step. The action handles PR comments automatically. This fixes error: "ENOENT: no such file or directory, open 'structure-results.md'" Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
The claude-code-action posts comments directly to PRs. The Post Review Comment step tried to read CLAUDE_OUTPUT env var which doesn't exist, resulting in unhelpful 'check logs' message. Removed the redundant post-comment step. The action handles PR comments automatically. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
📝 Documentation Review Summary
Review powered by Claude Code SME agents To apply fixes: Comment |
📝 Documentation Review Summary
Review powered by Claude Code SME agents To apply fixes: Comment |
| ### 1. Heading Analysis | ||
|
|
||
| Check for: | ||
| - Exactly one H1 (should be in frontmatter `title:` or first `#`) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we just standardize on one or the other? E.g., always use title: and never H1?
| ## Prerequisites | ||
|
|
||
| - Nextflow 23.04 or later | ||
| - A Seqera Platform workspace with Developer role or higher | ||
| - Your pipeline repository accessible to Seqera Platform | ||
|
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
My personal preference here is to use headings as headings, rather than introductions to points or steps. E.g.,
H2 Title
You need the following to get started:
- x
- y
- z
| - **Hyphens (-)**: For compound words and line breaks | ||
| - **Consistent spacing**: No spaces around em dashes, spaces around en dashes in ranges | ||
|
|
||
| ### Heading Punctuation |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Maybe add a point about sentence case here? I always seem to be fighting AI on this, but maybe I'm an edge case.
|
|
||
| Process files by: | ||
| 1. Parse markdown to identify different content types | ||
| 2. Skip code blocks and inline code |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Should a point about code blocks be added here? Check indentation? Make sure they are typed?
| - Abbreviations: X issues | ||
| ``` | ||
|
|
||
| ## Search Patterns |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I like how you've done this. I was going to suggest running this alongside vale to detect all of this, but this makes just as much sense.
christopher-hakkaart
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This all looks excellent. I've added a few minor comments, but they are extremely minor and more opinion-based. My view is that we should implement and iterate once we have some more data points of what it's doing well or poorly.
📝 Documentation Review Summary
Review powered by Claude Code SME agents To apply fixes: Comment |
No description provided.