Skip to content

chore: set up Matt Pocock engineering skills config#523

Closed
xdotli wants to merge 1 commit into
v0.5-integrationfrom
claude/setup-mattpocock-skills
Closed

chore: set up Matt Pocock engineering skills config#523
xdotli wants to merge 1 commit into
v0.5-integrationfrom
claude/setup-mattpocock-skills

Conversation

@xdotli
Copy link
Copy Markdown
Member

@xdotli xdotli commented May 25, 2026

Scaffolds the per-repo configuration the Matt Pocock engineering skills assume, so downstream skills (to-issues, triage, diagnose, tdd, improve-codebase-architecture, zoom-out) read this repo's context correctly.

What's added

  • AGENTS.md gains an ## Agent skills section pointing at three short docs. CLAUDE.md is left as the existing one-line pointer to AGENTS.md — no duplicate created (per the skill's hard rule).
  • docs/agents/issue-tracker.md — GitHub Issues via gh CLI (matches what we already do).
  • docs/agents/triage-labels.md — canonical vocabulary (needs-triage, needs-info, ready-for-agent, ready-for-human, wontfix). wontfix already exists in this repo; the file records the gh label create commands for the other four so they can be added when triage actually starts.
  • docs/agents/domain.md — confirms single-context layout (CONTEXT.md + docs/adr/ at the repo root). Neither file exists yet; producer skills create them lazily.

Decisions captured

Decision Choice
Issue tracker GitHub Issues at benchflow-ai/benchflow
Triage label vocabulary Canonical names verbatim
Domain docs layout Single-context

Optional follow-up

Run these to create the four new triage labels (skipped here because that's an external write on the repo settings):

gh label create needs-triage --description "Maintainer needs to evaluate" --color "fbca04" --repo benchflow-ai/benchflow
gh label create needs-info --description "Waiting on reporter" --color "d876e3" --repo benchflow-ai/benchflow
gh label create ready-for-agent --description "Fully specified, AFK-ready" --color "0e8a16" --repo benchflow-ai/benchflow
gh label create ready-for-human --description "Needs human implementation" --color "5319e7" --repo benchflow-ai/benchflow

Base branch note

Targeting v0.5-integration for consistency with the rest of this session's docs work; will flow into main when #344 lands. (AGENTS.md's trunk-based rule says branch off main — happy to retarget if you'd rather land this independently of the v0.5 release.)


Note

Low Risk
Documentation-only changes with no application code, auth, or data-path impact.

Overview
Adds per-repo agent configuration so Matt Pocock engineering skills can resolve this repo’s issue tracker, triage vocabulary, and domain-doc layout.

AGENTS.md now has an ## Agent skills section with pointers to three new docs under docs/agents/.

issue-tracker.md standardizes GitHub Issues on benchflow-ai/benchflow via gh (create/view/list/comment/label/close).

triage-labels.md maps canonical roles to label names (needs-triage, needs-info, ready-for-agent, ready-for-human, wontfix) and documents gh label create for the four labels not yet in the repo.

domain.md records a single-context layout (CONTEXT.md, docs/adr/ at repo root), instructs agents to read those when present, proceed silently if missing, use glossary terms, and flag ADR conflicts.

Reviewed by Cursor Bugbot for commit ede2f8a. Bugbot is set up for automated code reviews on this repo. Configure here.

Copy link
Copy Markdown

@cursor cursor Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Cursor Bugbot has reviewed your changes and found 1 potential issue.

Fix All in Cursor

❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.

Reviewed by Cursor Bugbot for commit 5e42083. Configure here.

def base64url_decode(s: str) -> bytes:
"""Decode base64url string to bytes."""
# Add padding
padded = s + "=" * (4 - len(s) % 4)
Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Incorrect base64url padding when length is multiple of 4

Low Severity

The padding formula "=" * (4 - len(s) % 4) adds 4 padding characters when len(s) is already a multiple of 4 (since 4 - 0 = 4). The correct formula is "=" * (-len(s) % 4) which adds 0 characters when no padding is needed. Python's base64.urlsafe_b64decode happens to tolerate extra padding, so this won't crash in practice, but the logic is incorrect.

Fix in Cursor Fix in Web

Reviewed by Cursor Bugbot for commit 5e42083. Configure here.

Copy link
Copy Markdown

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5e42083cb3

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment on lines +519 to +523
"--network",
f"{project_name}_default",
"--label",
f"com.docker.compose.project={project_name}",
"--label",
Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Reapply Compose mounts during snapshot restore

When restore() recreates main with a raw docker run, it only sets name/network/labels and never reattaches the Compose bind mounts. The baseline Compose spec mounts verifier logs, agent logs, and artifacts (src/benchflow/sandbox/_compose_files/docker-compose-base.yaml), so after a restore those paths are no longer host-backed and subsequent logging/artifact writes can disappear from the rollout workspace. This breaks branch rollouts that continue after restore because they lose expected filesystem wiring even though the container is running.

Useful? React with 👍 / 👎.

Scaffolds the per-repo configuration the engineering skills assume:

- AGENTS.md gains an `## Agent skills` block pointing at the three docs
- docs/agents/issue-tracker.md — GitHub Issues via `gh` CLI
- docs/agents/triage-labels.md — canonical vocabulary; only `wontfix`
  exists today, `gh label create` commands recorded for the other four
- docs/agents/domain.md — single-context (CONTEXT.md + docs/adr/ at root)

CLAUDE.md is left as the existing one-line pointer to AGENTS.md; no
duplicate created.

Spec: https://github.com/mattpocock/skills/blob/main/skills/engineering/setup-matt-pocock-skills/SKILL.md
@xdotli xdotli force-pushed the claude/setup-mattpocock-skills branch from 5e42083 to ede2f8a Compare May 25, 2026 05:00
@bingran-you bingran-you added the review:pending PR is ready-for-review, no reviewer engagement yet. label May 25, 2026
@bingran-you bingran-you deleted the branch v0.5-integration May 26, 2026 19:34
@bingran-you bingran-you deleted the claude/setup-mattpocock-skills branch May 26, 2026 20:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

review:pending PR is ready-for-review, no reviewer engagement yet.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants