Fix agentry CI workflows and add feature-implement#120
Conversation
…plement All CI workflows now pass --output-format json, --input, and --binder github-actions flags matching the hand-tuned patterns in the agentry reference repo. Adds feature-implement workflow triggered by the `feature` label. Replaces standalone triage with the planning pipeline (triage + task decomposition) on issue open.
📝 WalkthroughWalkthroughUpdated several Agentry GitHub Actions workflows to pin action versions, add concurrency controls, and require JSON output with explicit inputs. Added a new "Agentry: Feature Implement" workflow plus its agent workflow and prompt for automated feature implementation triggered by issue labels. Changes
Sequence DiagramsequenceDiagram
participant Issue as GitHub Issue
participant GHA as GitHub Actions
participant Agent as Claude-Code Agent
participant Repo as Repository
participant Output as Result Store
Issue->>GHA: labeled event (label = feature)
GHA->>GHA: checkout, setup Python/npm, install agentry
GHA->>Agent: run workflows/feature-implement.yaml (--output-format json, repository-ref=.)
Agent->>Repo: read issue description and repository contents
Agent->>Agent: decide: implement directly or decompose
alt Direct implement
Agent->>Repo: read/modify role files, update molecule scenarios
Agent->>Repo: run shell commands, commit changes
Agent->>Repo: create pull request and link to issue
else Decompose
Agent->>Agent: create sub-tasks
Agent->>Repo: create and label sub-issues, link to parent
end
Agent->>Output: write result JSON (action, reasoning, pr_url/sub_issues)
Output->>Issue: post summary comment / link
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes 🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In @.github/workflows/agentry-feature-implement.yaml:
- Around line 10-13: The agentry job lacks a per-issue concurrency group causing
overlapping labeled events to produce duplicate write-side effects; add a
concurrency block to the agentry job (near the job header where "agentry:" and
"runs-on: ubuntu-latest" are defined) that sets a unique group per issue/PR (for
example use an expression like agentry-${{ github.event.issue.number ||
github.event.pull_request.number || github.ref }}) and set cancel-in-progress:
true so concurrent runs for the same issue/PR are serialized/cancelled to
prevent duplicate PR/comment races.
🪄 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: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 435d72bb-4281-4f6f-be0d-8e997bcd544d
📒 Files selected for processing (6)
.github/workflows/agentry-bug-fix.yaml.github/workflows/agentry-code-review.yaml.github/workflows/agentry-feature-implement.yaml.github/workflows/agentry-triage.yamlworkflows/feature-implement.yamlworkflows/prompts/feature-implement.md
Pin actions/checkout and actions/setup-python to full commit SHAs matching the rest of the repo's workflows. Add per-issue/PR concurrency groups to all agentry workflows to prevent duplicate runs from racing. Document why contents: write is needed on bug-fix and feature-implement.
There was a problem hiding this comment.
🧹 Nitpick comments (1)
.github/workflows/agentry-code-review.yaml (1)
24-27: Pin CLI install sources to immutable versions for reproducible CI.
@anthropic-ai/claude-codeandagentryare installed from mutable references, which can change behavior between runs. Pin a specific package version for the npm package and a git commit SHA for agentry. This pattern appears in.github/workflows/agentry-code-review.yaml(lines 25, 27),.github/workflows/agentry-feature-implement.yaml(lines 26, 28),.github/workflows/agentry-bug-fix.yaml(lines 26, 28), and.github/workflows/agentry-triage.yaml(lines 22, 24).Suggested patch
- - name: Install Claude Code - run: npm install -g `@anthropic-ai/claude-code` + - name: Install Claude Code + run: npm install -g `@anthropic-ai/claude-code`@<pinned-version> - - name: Install agentry - run: pip install "agentry @ git+https://github.com/norrietaylor/agentry.git" + - name: Install agentry + run: pip install "agentry @ git+https://github.com/norrietaylor/agentry.git@<commit-sha>"🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.github/workflows/agentry-code-review.yaml around lines 24 - 27, Update the workflow steps that install the CLIs so they use immutable versions: change the npm install step that runs "@anthropic-ai/claude-code" (step name "Install Claude Code") to install a specific released version (e.g., npm install -g `@anthropic-ai/claude-code`@<version>) and change the pip install step that installs "agentry" (step name "Install agentry") to point to a fixed git commit SHA (e.g., pip install "agentry @ git+https://github.com/norrietaylor/agentry.git@<commit-sha>"). Apply the same pinning pattern to each workflow that contains these steps (the other workflow files with the same "Install Claude Code" and "Install agentry" steps).
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In @.github/workflows/agentry-code-review.yaml:
- Around line 24-27: Update the workflow steps that install the CLIs so they use
immutable versions: change the npm install step that runs
"@anthropic-ai/claude-code" (step name "Install Claude Code") to install a
specific released version (e.g., npm install -g
`@anthropic-ai/claude-code`@<version>) and change the pip install step that
installs "agentry" (step name "Install agentry") to point to a fixed git commit
SHA (e.g., pip install "agentry @
git+https://github.com/norrietaylor/agentry.git@<commit-sha>"). Apply the same
pinning pattern to each workflow that contains these steps (the other workflow
files with the same "Install Claude Code" and "Install agentry" steps).
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 30e247c4-105b-41ac-96a0-c6426eb4a268
📒 Files selected for processing (4)
.github/workflows/agentry-bug-fix.yaml.github/workflows/agentry-code-review.yaml.github/workflows/agentry-feature-implement.yaml.github/workflows/agentry-triage.yaml
🚧 Files skipped from review as they are similar to previous changes (2)
- .github/workflows/agentry-triage.yaml
- .github/workflows/agentry-bug-fix.yaml
Fixes the agentry workflows that were failing because the generated CI templates
were used as-is without the manual tuning that agentry expects. All workflows now
pass --output-format json, --input, and --binder github-actions flags.
Also adds the feature-implement workflow (triggered by
featurelabel) andreplaces the standalone triage with the full planning pipeline on issue open.
Summary by CodeRabbit
New Features
Chores