agentleFS
encoreshaoencoreshao/encore-skills/skills/eng-workflow/SKILL.md

eng-workflow

Skill2 starsChanged 15 days agoMITView on GitHub

Full GitLab development loop — from issue to confirmed-resolved. Covers write-issue, analyze-issue, fix-issue, review-code, create-mr, summarize-issue, triage-issue, and post-merge verification.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 7 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use write-issue. Focus on the root cause, not just the symptom.

    line 40
  2. Very specific:

    Use analyze-issue. The goal is to understand the root cause — not just read the ticket and start coding.

    line 48
  3. Distinctive:

    Gate: All acceptance criteria met.

    line 58
  4. Very specific:

    Use review-code. Start by asking: does this actually solve the problem from the issue?

    line 64
  5. Very specific:

    If the user asks for a recap on the issue itself (beyond the brief comment create-mr already posts), use summarize-issue.

    line 80
  6. Distinctive:

    Check the target environment (staging or production): does the original problem still occur?

    line 88
  7. Distinctive:

    If you can reproduce the original issue — do it now and confirm it no longer happens

    line 89
  8. Distinctive:

    If the issue isn't auto-closed, close it manually with a comment confirming resolution

    line 90
  9. Distinctive:

    If post-merge verification reveals the fix didn't work: reopen the issue, note what was tried and why it didn't work, and start the loop …

    line 97
  10. Distinctive:

    The issue defines done: When in doubt about whether something is complete, re-read the issue.

    line 102
  11. Distinctive:

    Small issues, small MRs: If scope grows, open a new issue.

    line 103

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Engineer Workflow

Full development loop. The goal is not to merge an MR — it's to confirm the problem is actually gone.

The loop

write-issue → analyze-issue → fix-issue → review-code → create-mr → [merge]
      ↑                                                                 |
      └──────────────────────── new issue from feedback ────────────────┘

Entry points

Where you areStart here
Have a rough idea or bug reportwrite-issue
Have a GitLab issueanalyze-issue
Have an analysis, ready to codefix-issue
Code done, ready to shipreview-code
Issue fixed, MR open, want a summary posted to the issuesummarize-issue
Issue has comments that may need your replytriage-issue

Phase guide

Phase 0: Write Issue

Skip if issue already exists in GitLab.

Use write-issue. Focus on the root cause, not just the symptom. Define what "fixed" looks like before any code is written.

Gate: Issue has a clear problem statement and testable acceptance criteria.


Phase 1: Analyze Issue

Use analyze-issue. The goal is to understand the root cause — not just read the ticket and start coding.

Gate: Root cause identified. No open blockers. Approach defined.


Phase 2: Fix Issue

Use fix-issue. If you're on a protected branch (main/master/develop/staging), it creates <type>/<issue-number>-<func-name> and records the branch you came from as the MR base. Read the existing code before touching it. Fix the root cause, not the symptom. Verify the problem is gone before moving on.

Gate: All acceptance criteria met. Problem confirmed gone — not just tests green.


Phase 3: Review Code

Use review-code. Start by asking: does this actually solve the problem from the issue? Then check security, correctness, simplicity.

Gate: Verdict is "Ready to merge". No open ❌ items.


Phase 4: Create MR

Use create-mr. It targets the branch you branched from (recorded in Phase 2), not always main. Write a description that tells reviewers what problem was solved and whether it's fixed — in plain language, not a file list. create-mr also posts a brief key-result comment on the issue; an MR isn't done until that comment exists.

Gate: MR is open, targets the right base branch, CI is passing, description links to the issue.


Optional: Summarize on the issue

If the user asks for a recap on the issue itself (beyond the brief comment create-mr already posts), use summarize-issue. High-level, markdown, no diff dump — written for someone who won't read the code.


Phase 5: After merge — verify in the real environment

This step is skipped too often. Don't skip it.

  • Check the target environment (staging or production): does the original problem still occur?
  • If you can reproduce the original issue — do it now and confirm it no longer happens
  • If the issue isn't auto-closed, close it manually with a comment confirming resolution
glab issue close <number>
glab issue note <number> --message "Confirmed resolved in <env> after merge."

If post-merge verification reveals the fix didn't work: reopen the issue, note what was tried and why it didn't work, and start the loop again from Phase 1 with better information.

Principles

  • Fix the root cause, not the symptom. A patch that hides the problem is not a fix.
  • The issue defines done. When in doubt about whether something is complete, re-read the issue.
  • Small issues, small MRs. If scope grows, open a new issue. Don't expand the MR.
  • Looping is normal. Finding a problem in Phase 3 or 4 is better than finding it in production.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.