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 jontsai/claude-plugins --skill mem-rescuegit clone --depth 1 https://github.com/jontsai/claude-pluginsWrote 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/jontsai/claude-plugins/mem-rescue)<a href="https://agentmods.dev/skills/jontsai/claude-plugins/mem-rescue"><img src="https://agentmods.dev/badge/skills/jontsai/claude-plugins/mem-rescue/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/jontsai/claude-plugins/mem-rescue"><img src="https://agentmods.dev/badge/skills/jontsai/claude-plugins/mem-rescue.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 | $0.00143 | $0.00984 |
| Opus 5.5 | $0.00057 | $0.00394 |
| Sonnet 5.5 | $0.00029 | $0.00197 |
| Haiku 4.5 | $0.00014 | $0.00098 |
Grade A, and why
mem-rescue 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 today.
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 — 70 lines — stays where its author put it; the contents beside it link to each section on GitHub.
mem-rescue
Two bundled scripts replace the dozen ad-hoc top, ps, and vm_stat calls a memory
diagnosis normally takes. Both live in this skill's scripts/ directory; run them from the
directory this skill was loaded from (the loader tells you where that is). Running them by
path, with no shell pipeline around them, also lets the user allowlist them once in their
agent so they never prompt.
1. Diagnose (read-only, always safe)
scripts/memcheck.sh # human report
scripts/memcheck.sh --json # same data, machine-readable
Output: a verdict (OK / STRAINED / CRITICAL), the headline numbers, memory by app family, Chromium/Electron process counts, the largest renderer processes, leak suspects, crash reports from the last 24h, and ranked restart actions with estimated recoverable memory.
Run it once, read it, and go straight to the answer. Don't re-derive the numbers with your own
top or ps calls unless the report is missing something specific.
2. Rescue (quits and relaunches apps; confirm first)
scripts/rescue.sh --dry-run brave slack # shows what would happen
scripts/rescue.sh brave slack # does it
scripts/rescue.sh all # brave slack notion calendar
Quits are graceful, so Chromium browsers restore their tabs lazily and Slack/Notion reopen
where they were. The script never force-kills unless given --force, and it reports swap
before and after.
Quitting the user's browser or chat client interrupts their work. Unless the user already told you to restart things ("restart brave", "go ahead and rescue it"), show the ranked actions from the report and ask which ones to run. When they say yes, run rescue once, then run memcheck again and show the before/after.
Reading the report
- Verdict CRITICAL means swap is mostly full or free memory is under 20%. New processes (browser tabs, renderers) fail to launch because the compressor and swap have nowhere to put pages. The fix is "shed the biggest app family", not "close a few things".
- Demand is physical used plus swap used. When demand is far above installed RAM, the workload simply doesn't fit, and restarting the top app is the only quick lever.
- High CMPRS (or SWAP) relative to TOTAL on a browser means most tabs are idle background
tabs that have been compressed or swapped out. A restart recovers nearly all of it because
tabs come back as unloaded placeholders. Suggest enabling the browser's Memory Saver
(
brave://settings/system,chrome://settings/performance) so it discards idle tabs itself. - Leak suspects are single processes over 1 GB that have run for 2+ days. Electron renderers (Slack, Notion, Discord) grow unbounded; a relaunch resets them to a few hundred MB.
- Many swapfiles after days of uptime means the compressor is fragmented. macOS does shrink swap once memory is freed, so a reboot is a "when convenient" item, not an emergency.
- Crash reports for the browser at the time the user reported tab failures confirm memory
starvation rather than a browser bug. On macOS the
.ipsfile'sparentProcfield shows who launched the crashed process; if it was an automation agent, say so.
What ships with it
2 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.
- today First seen · 70 lines · 143 tokens per session scan A 3cbeef78587d
mem-rescue is a skill published in the GitHub repository jontsai/claude-plugins (6 stars, last pushed today), licensed MIT. It adds 143 tokens to every session and 984 once invoked, about $0.0006 per session on Opus 5.5. 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 2026-10-17.
Other skills, from other repositories
visual-explainer
Generate self-contained HTML pages that visually explain technical material — systems, code changes, plans, data. Use when the user asks for any diagram or visual explanation, when they want a slide deck or a longform document (user guide, how-to, reference page), when another skill needs one rendered, or proactively…
reflect-solution
Summarize WHAT WAS SHIPPED in a Claude Code session as a self-contained HTML slide deck — the elevator pitch (problem, solution, decisions, artifacts, verification, gaps), not a play-by-play of tool calls or debugging detours. Defaults to the current in-context session; optionally accepts a session ID or JSONL path.…
structured-plan-mode
This skill should be used when planning and tracking complex feature implementations that require systematic task decomposition. Use this skill to break down large features into manageable, well-documented tasks with clear dependencies, action items, and success criteria. The skill provides a structured template and…
glab
Expert guidance for using the GitLab CLI (glab) to manage GitLab issues, merge requests, CI/CD pipelines, repositories, and other GitLab operations from the command line. Use this skill when the user needs to interact with GitLab resources or perform GitLab workflows.
reflect-tree
Visualize a Claude Code session as a quest/skill tree — a navigable SVG graph where nodes are turns and edges show flow, with distinct visual encoding for normal flow, dead-ends, corrections, retries, reversals, and backtracking. Sibling to /reflect (which produces an incidents+recommendations dashboard); this one…
update-component-reference
This skill should be used when the user wants to add components (commands, agents, skills, hooks, or MCP servers) to the Component Reference section of the website.