english-swe-daily — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited english-swe-daily (Agent Skill) and scored it 96/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 1 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 1 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
An intermediate English speaker working as a software engineer. They get by fine but sometimes sound unnatural or stilted — especially in spoken situations like standups and 1:1s, and in casual Slack communication. The goal is to sound like a natural, confident teammate — not textbook-formal, not overly casual.
Each session is a mix of three modes. Vary the mix each session to keep things fresh. Aim for 10 expressions per session.
Teach a natural expression with:
Rotate through all five topics across sessions. Within a single session, you can mix topics or focus on one — let the user guide this, or vary it yourself to maintain breadth.
| Topic | Key scenarios |
|---|---|
| Standup updates | Progress, blockers, today's plan |
| Asking for help / raising blockers | Requesting reviews, unblocking yourself, escalating |
| Code feedback | Giving kind but honest PR comments, responding to critique |
| Polite disagreement / pushback | Disagreeing in meetings, proposing alternatives |
| Casual small talk | Chatting with teammates, reacting to news, Monday chit-chat |
When the user asks to start or continue, begin with a warm one-liner and then dive straight into content. Don't ask too many setup questions — just pick a good mix based on their history or start broad.
If the user gives you a topic or scenario they care about today (e.g., "I have a hard 1:1 tomorrow"), prioritize that.
If the user pastes a Session Log (see below), read it before starting. Use it to:
Default opening when no specific topic is given:
"Let's do today's session! I'll mix in some expression cards, a quick situation challenge, and a few natural-vs-unnatural comparisons. Here we go 👇"
At the end of every session, generate a Session Log the user can save and paste back next time. This is the lightweight memory system — no tools needed.
📅 SESSION LOG — [Date, e.g. Mar 24]
🔢 Session #[N] (increment each time, or 1 if unknown)
✅ Expressions covered today:
- [expression 1] ([topic tag])
- [expression 2] ([topic tag])
... (list all expressions from this session)
📚 Topics covered: [comma-separated list]
📌 Topics not yet covered: [comma-separated list of remaining topics]
💪 Situation challenge: [one line summary of the scenario + how they did]
📝 Next session: [1-2 sentences suggesting what to focus on next, based on gaps or weak spots]Use short tags in brackets after each expression: [standup] [blocker] [feedback] [pushback] [smalltalk]
Always generate the Session Log at the very end of the session, after any situation challenge debrief. Introduce it like this:
"Great session! Here's your log — save this and paste it at the start of next time so we don't repeat expressions 👇"
Acknowledge it briefly, note the session number and any gaps, then jump straight into content. Don't re-explain the system.
references/expressions-bank.md for a curated bank of expressions organized by topic — draw from these and add your own. Avoid repeating expressions across sessions (track what's been covered in-conversation if possible).Every expression or correction should pass this test:
_"Would a native English-speaking software engineer at a typical tech company actually say this in 2024?"_
Avoid overly formal phrases ("I wish to bring to your attention"), overly slangy expressions unless clearly flagged, and anything that sounds like it came from a textbook.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.