General-purpose coding policy for Baruch's AI agents
73
91%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
View guide
Low
Low-risk findings worth noting
"""The skills that push a branch to origin open its PR in origin's repository.
`release` Step 2 and `onboard-repo` Step 8 push with `git push ... origin` and
then run `gh pr create`. A bare `gh pr create` targets `gh repo set-default`,
so a fork whose default is its upstream parent could open the PR in a
repository the operator does not own (#666).
"""
import re
import unittest
from pathlib import Path
SKILLS = Path(__file__).resolve().parents[2]
#: (skill, step number) pairs whose step opens a PR for a branch pushed to origin.
PR_STEPS = (("release", 2), ("onboard-repo", 8))
COMMAND = re.compile(r"`(gh pr create[^`]*)`")
RESOLVE = 'python3 "$CP/skills/release/origin-repo.py" .'
def step(text, number):
"""The body of `## Step <number> — ...`, up to the next H2."""
match = re.search(r"^## Step {} —.*?(?=^## |\Z)".format(number), text, re.S | re.M)
if match is None:
raise AssertionError("no Step {}".format(number))
return match.group(0)
class PrCreateRepoBindingTest(unittest.TestCase):
def test_each_pr_step_resolves_origin_then_names_it(self):
for skill, number in PR_STEPS:
with self.subTest(skill=skill):
# Whitespace-normalized: a Markdown line wrap never changes an instruction.
body = " ".join(step((SKILLS / skill / "SKILL.md").read_text(encoding="utf-8"), number).split())
found = COMMAND.findall(body)
self.assertTrue(found, "{} Step {} has no gh pr create command".format(skill, number))
self.assertIn(RESOLVE, body)
for cmd in found:
self.assertIn("--repo <repo>", cmd)
self.assertLess(body.index(RESOLVE), body.index(cmd))
if __name__ == "__main__":
unittest.main().tessl-plugin
hooks
rules
skills
adopt-fork-pr
herdr-foreman
classify
foreman
references
templates
tests
herdr-standup
migrate-to-plugin
onboard-repo
release
references
tests