Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/maxence2997/mx-harness/mx-prnpx skills add maxence2997/mx-harness --skill mx-prgit clone --depth 1 https://github.com/maxence2997/mx-harnessWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00066 | $0.02426 |
| Opus 5 | $0.00033 | $0.01213 |
| Sonnet 5 | $0.00013 | $0.00485 |
| Haiku 4.5 | $0.00007 | $0.00243 |
Grade A, and why
mx-pr scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 294 lines — stays where its author put it; the contents beside it link to each section on GitHub.
mx-pr
Trigger
/mx-pr <name> ← draft PR for named feature
/mx-pr ← infer from active spec or ask
Orchestrated mode
This skill has interactive pauses (Steps 4 and 5). When invoked from an orchestrator that declares auto-proceed for its PR gate (e.g. mx-flow GATE 4), the orchestrator's gate table overrides those pauses: still display everything you would have shown, but proceed without waiting — pause only if you cannot determine how to proceed (no remote, ambiguous platform, missing credentials). When invoked directly by the user, the pauses apply as written.
Because auto-proceed removes the human review of the draft, add one check in its place: if a subagent tool (Agent/Task) is available, spawn a fresh read-back agent on the draft before Step 6 — criteria: every factual claim traces to spec.md or the git log; every referenced issue number exists; the body names no local-only files (Step 3's committed-files-only rule). Fix findings before pushing. If no subagent tool exists, re-check the draft yourself against those criteria and label the result "self-checked, single-context" in the output.
Path resolution
Resolve two base directories before any file operation:
REPO_ROOT=$(git rev-parse --show-toplevel)
PROJECT=$(basename "$REPO_ROOT")
| Variable | Path | Used for |
|---|---|---|
GLOBAL_MX |
~/.mx/<project>/<name>/ |
Reading spec.md (permanent) |
LOCAL_MX |
<repo-root>/.mx/<name>/ |
Writing PR drafts to tmp/ (ephemeral) |
Step 1 — Gather context
Read GLOBAL_MX/spec.md (~/.mx/<project>/<name>/spec.md) for the
What/Why/How summary.
Resolve the base branch once — every later step uses this value (the log below, the content check in Step 2, and the PR target in Step 6):
if git rev-parse --verify --quiet develop >/dev/null || git rev-parse --verify --quiet origin/develop >/dev/null; then
BASE_BRANCH=develop
elif git rev-parse --verify --quiet main >/dev/null || git rev-parse --verify --quiet origin/main >/dev/null; then
BASE_BRANCH=main
else
BASE_BRANCH="" # neither exists — ask the user, do not guess
fi
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 294 lines · 66 tokens per session scan A 8cadf14af936
mx-pr is a skill published in the GitHub repository maxence2997/mx-harness (5 stars, last pushed 6d ago), licensed MIT. It adds 66 tokens to every session and 2,426 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
git-workflow-and-versioning
Structures git workflow practices. Use when making any code change. Use when committing, branching, resolving conflicts, opening or reviewing a pull request (PR), pushing to a remote, or when you need to organize work across multiple parallel streams. Use when cutting a release, choosing a semantic version bump…
comet-safe-delivery
在保护无关脏改动、关联 worktree、子模块和用户明确边界的前提下，提交、推送、合并或完成范围明确的 Comet 变更。用户要求提交、推送、合并回目标分支、清理 worktree 或交付已准备好的改动时使用。.
changelog-composer
Generates structured changelogs and release notes from git history and PRs, classifying breaking changes, features, fixes, performance, docs. Triggers on: "generate changelog", "write release notes", "what changed since", "prepare release", "release notes for", "diff since tag".
github
GitHub CLI operations via gh for issues, PRs, Actions, releases, and REST/GraphQL API with --json/--jq parsing. Triggers on: "create an issue", "submit a PR", "check CI status", "why did CI fail", "merge a PR", or pasted GitHub URLs.
conventional-commits
Enforce, validate, lint, and manage Conventional Commits across git repositories. Generates semantic changelogs and installs git hooks. Use this skill whenever the user mentions commit messages, commit conventions, conventional commits, semantic versioning, changelogs, commit linting, git hooks for commits, or wants…
git-workflow
Use for branching, committing, history repair, and release hygiene. Covers atomic commits, rebase versus merge, bisect, reflog recovery, and undoing mistakes safely.