agentleFS
skforecastskforecast/skforecast-ai/.claude/skills/open-pr/SKILL.md

open-pr

Skill52 starsChanged 47 days agoApache-2.0View on GitHub

Prepare and open a skforecast-ai pull request against the current release branch, after running the tests and checks that CI does not run on release branches.

  • Commits and pushes

67 other projects use a version of this, 1 of them edited.

Adopted by

Most recently taken up by

Used with

Projects with it also use review-change, commit-push and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • thlaureSkill

    review-change

    Use this skill when the user asks for a code review, a correctness pass, a check before PR, or asks whether a Signalist change follows project conventions.

    Also in 4 projects that use this

  • DataCoveySkill

    commit-push

    Commit and push the changes made on this chat session to the current branch. Use best practices from Conventional Commits.

    Also in 3 projects that use this

  • Codagent-AISkill

    implement-change

    Implements an approved change in one continuous session, with tiered test runs. Use when asked to implement a change from its approved definition, or to invoke codagent:implement-change.

    Also in 3 projects that use this

  • saffron-healthSkill

    address-review

    Address PR review comments systematically. Use when responding to code review feedback on a pull request.

    Also in 3 projects that use this

  • DockerSkill

    review-pr

    Review one or more incoming Docker documentation pull requests as a maintainer. Independently validate technical claims, assess editorial fit and information architecture, choose a verdict, and draft exact inline or PR-wide feedback behind a confirmation gate. Use for requests such as "review PR 123", "is this PR correct?", "does this information belong here?", "validate this PR", or "help review backlog PRs". Do not use to maintain or fix a PR you own; use maintain-pr for that.

    Also in 10 projects that use this

  • SentrySkill

    write-tests

    Write tests following project conventions. Use when adding new tests or modifying existing ones.

    Also in 4 projects that use this

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 1 is distinctive, and 6 are very specific to this project.

  • 0

    Common

    Most of the 66 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Very specific:

    If the current branch is main or the release branch, create <type>/<slug> from BASE first (type feature, fix, docs, chore or perf, the …

    line 20
  2. Very specific:

    If the branch is behind BASE, say so.

    line 32
  3. Very specific:

    Run /verify on the whole branch: its scope is git diff --name-only BASE...HEAD plus the working tree, not only the last commit.

    line 46
  4. Very specific:

    If the branch touches the paths of the AI context check, run what CI runs, with the Python command of /verify (the second one needs network …

    line 49
  5. Very specific:

    If the branch changes what the LLM receives (llm/context.py, llm/prompts.py, skforecast_ai/skills/, recommendation/explanation.py, …

    line 59
  6. Distinctive:

    Write the body to a file in the scratchpad:

    line 91
  7. Very specific:

    Add the link of the new PR to its release note entries, (#123) at the end of each one (see the trailers in /release-note), then commit and …

    line 112

“Common” is measured: kept by at least 6 in 10 of the 66 edited versions of this file we could read. “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

Open a pull request

Title hint from the user: $ARGUMENTS

1. Branches

  • Release branch: read __version__ in skforecast_ai/__init__.py and derive X.Y.x (0.4.0 -> 0.4.x). On main, which keeps the previous release, it is the highest X.Y.x on origin instead. The PR base is that branch, never main.
  • git fetch origin and compare against origin/<release-branch>: the local copy of the release branch may be stale. Below, BASE stands for origin/<release-branch>.
  • If the current branch is main or the release branch, create <type>/<slug> from BASE first (type feature, fix, docs, chore or perf, the names the cloud hook accepts). If it is a claude/... (cloud session) or worktree-... branch, rename it with git branch -m <type>/<slug> after checking that git log --oneline BASE..HEAD lists only the commits of this work: a pull request keeps the name of its branch, and the Bash guard blocks pushing those.
  • Review git log --oneline BASE..HEAD, git diff BASE...HEAD --stat and git status. Commit the uncommitted changes that belong to the PR without asking; leave the rest out (see the dev/ files in step 2) and list them in the report.
  • If the branch is behind BASE, say so; do not rebase or merge without asking.
  • gh pr view tells whether the branch already has a PR. If it does, update that one (push, then gh pr edit if the body must change) instead of opening a second one.

2. Checks

CI runs lint and the unit tests only on pull requests to main. On a release branch it only runs the AI context check, and only when skforecast_ai/skills/, skforecast_ai/resources/ or skforecast_ai/llm/prompts.py change. The checks run here are the only ones before merge.

  1. Run /verify on the whole branch: its scope is git diff --name-only BASE...HEAD plus the working tree, not only the last commit. Stop on failures.
  2. If the branch touches the paths of the AI context check, run what CI runs, with the Python command of /verify (the second one needs network access):
    python tools/ai/measure_skill_tokens.py --check
    python tools/ai/sync_skforecast_assets.py --check --branch auto
    Changes to skforecast_ai/skills/ or skforecast_ai/resources/llms-base.txt must come from the sync script, never from a hand edit.
  3. If the branch changes what the LLM receives (llm/context.py, llm/prompts.py, skforecast_ai/skills/, recommendation/explanation.py, execution/backtesting_runner.py, execution/comparison.py), the goldens must be regenerated on purpose (/llm-context-change). The real-model run of tools/ai/check_ask_context.py costs money, so the user launches it: ask whether it was run and whether its report is saved under tools/ai/ask_context_reports/; otherwise list it as pending in the PR body.
  4. User-facing changes (API, CLI output, generated scripts, warnings) have an entry in docs/releases/releases.md under the version in development, as short entries (one or two sentences; the detail goes in the PR description). If one is missing or over the length ceiling (python tools/docs/check_release_notes.py), follow /release-note and show the result to the user.
  5. dev/ holds working notes. No dev/handoff_*.md file (written by /handoff) is part of the PR: delete it first, asking the user. The other notes are versioned only on purpose (dev/roadmap-0.6.0.md): list the dev/ files in the diff and ask whether each belongs to the PR.

3. Open the PR

  • GitHub account: git remote -v shows which account pushes (an SSH host alias may point to a different key than gh uses), and gh api user --jq .login shows which account gh would open the PR with. If they are not the same person, stop and ask: the user can switch with gh auth switch (or log in the other account with gh auth login), or open the PR from the link printed by git push. Never switch accounts yourself.
  • Push the branch with git push -u origin <branch> (a local session asks for confirmation; the Bash guard blocks pushes to main).
  • Write the body to a file in the scratchpad:
    • Maintainers (Javier Escobar Ortiz, Joaquin Amat Rodrigo, as in pyproject.toml; compare with git config user.name): a ## Description of what changes and why (link the issue if any) and a ## How was this tested? section listing each check run, with its command and result (number of tests passed), plus what was not run (for example the real-model check_ask_context.py run). Leave out the checklist of .github/pull_request_template.md: it is written for external contributors (the CLA does not apply to the copyright holders), and the results say more than ticked boxes. The ## Description carries the technical detail that the release note leaves out: the cause of a bug, how it was fixed, the measurements and the edge cases.
    • Anyone else: fill .github/pull_request_template.md with the Description, how it was tested and the checklist, ticking only what was actually verified.
  • Create it with gh pr create --base <release-branch> --title "<title>" --body-file <file>. The title follows the commit style of the repository (fix(forecast): ..., docs(dev): ...). In a cloud session, use the session's PR flow if gh is not authenticated.
  • Add the link of the new PR to its release note entries, ([#123](https://github.com/skforecast/skforecast-ai/pull/123)) at the end of each one (see the trailers in /release-note), then commit and push that change to the same branch.
  • The title, body and commits carry only the author's identity: no AI co-author trailer, session link, "Generated with" or "Generated by" line or mention of Claude (the Bash guard blocks them, also in a --body-file, and so does the guard of the GitHub MCP tools). In a cloud session the GitHub MCP server appends a "Generated by Claude Code" footer after the call: read the new pull request back and remove it with the update tool (a hook reminds you). No en dashes or em dashes.

4. Report

Give the PR URL, the checks run with their results, and anything left for the reviewer (pending real-model run, dev/ files left out, open questions).

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.