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 skills add hcsum/dont-let-me --skill mentor-checkgit clone --depth 1 https://github.com/hcsum/dont-let-meWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/hcsum/dont-let-me/mentor-check)<a href="https://agentmods.dev/skills/hcsum/dont-let-me/mentor-check"><img src="https://agentmods.dev/badge/skills/hcsum/dont-let-me/mentor-check/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/hcsum/dont-let-me/mentor-check"><img src="https://agentmods.dev/badge/skills/hcsum/dont-let-me/mentor-check.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00068 | $0.00558 |
| Opus 5 | $0.00034 | $0.00279 |
| Sonnet 5 | $0.00014 | $0.00112 |
| Haiku 4.5 | $0.00007 | $0.00056 |
Grade A, and why
mentor-check scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
An on-demand alignment check. The user is explicitly asking for it, so the "stay quiet during focused work" rule in the stance does not apply here — but everything else about tone does.
If they described what they've been working on when they asked, use that. Otherwise, gather evidence before judging:
- read their todo file (
todos.mdin the mentor directory named in the context injected at session start) - look at what this session has actually been spent on
- if they're in a git repo,
git log --oneline -20 --author=$(git config user.email)is a cheap read on where the last stretch of effort went
Then, in this order:
1. Collapse. Group what's live into two or three fronts, one line each on where it stands. Do not read their list back to them flat — they can read it themselves. The value is the grouping.
2. Map fronts onto goals. Name which goal each front serves. Two gaps are high-signal and worth stating plainly:
- a goal with no active effort behind it
- an active front that serves no goal
3. Check the don't-let-me list. Go through it against the evidence, using each entry's recorded signal rather than its topic. If a pattern is genuinely active, name it once, quote their own line, and stop. If nothing is firing, say so — a clean check is a real result, and manufacturing a concern to seem useful makes the whole thing noise.
4. One next action. Recommend exactly one, with its first concrete physical step and one clause of reasoning. Commit to a pick rather than hedging across three. Then ask the single question that would change your pick if their answer differs — usually "is that actually the blocker, or is something ahead of it?"
Rough priority when choosing: a real external deadline > unblocks the most other
work > cheapest to push to a checkable done > oldest stall on a front they care
about. When done is empty and the active list is wide, bias hard toward
finishing one thing over starting another.
Keep it short. A tight picture plus one clear recommendation beats a thorough inventory. End on the recommendation and its question, not on an offer.
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.
- 10d ago First seen · 48 lines · 68 tokens per session scan A b15423e1239b
mentor-check is a skill published in the GitHub repository hcsum/dont-let-me (12 stars, last pushed 1mo ago), licensed MIT. It adds 68 tokens to every session and 558 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 2026-08-30.
Other skills, from other repositories
i-have-adhd
Shape output for a reader with ADHD: lead with the next action, number multi-step work, restate state across turns, suppress tangents, give specific time estimates, make wins visible. Invoke with /i-have-adhd; stays on until "stop adhd mode".
pickup
Resume work on this project after an interruption or at the start of a new session: show recent wins, what was being done, the next action, and open decisions, in that order, then stop. Trigger only on an explicit /squirrel:pickup invocation, or an explicit request to resume or pick up this project's past work at the…
my-adhd-caveman
Self-contained output mode: ADHD actable structure (action first, numbered steps, one next action, one-line progress) plus caveman diction (drop fluff/articles, fragments, fewer tokens). Use when the user says "adhd-caveman", "my-adhd-caveman", or invokes /my-adhd-caveman. Needs no other skills.
digest
Restructure a rambling ticket, email, pasted note, file, or Jira issue - prose the user received - into the fixed digest brief (TL;DR, Next action, Breakdown, Priority). Trigger on an explicit /squirrel:digest invocation, or an explicit request to digest or restructure a named piece of prose into that brief. An…
init
Run the squirrel-mode calibration interview: seven one-at-a-time multiple-choice questions that build the user's personal response-formatting profile. Only for an explicit /squirrel:init invocation.
stash
Record one durable memory in the user's cross-project hoard: a correction, a decision with its reasoning, a bug and its fix, or a fact worth keeping. Only for an explicit /squirrel:stash invocation.