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/deepgram/wsh/corenpx skills add deepgram/wsh --skill coregit clone --depth 1 https://github.com/deepgram/wshWhat 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 | $0.00070 | $0.05418 |
| Opus 5 | $0.00035 | $0.02709 |
| Sonnet 5 | $0.00014 | $0.01084 |
| Haiku 4.5 | $0.00007 | $0.00542 |
Grade B, and why
core scanned grade B with 2 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.
Sends data to an external URLmediumData exfiltration
A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.
curl -s -X POST --unix-socket $WSH_SOCK http://localhost/sessions/work/input -d $'ls -la\n' Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
MCP tools. Contains the complete HTTP API reference with working curl How it starts
The opening of the file, as written. The whole thing — 515 lines — stays where its author put it; the contents beside it link to each section on GitHub.
wsh: Terminal as a Service
You have access to wsh, an API that gives you direct control over
terminal sessions. You can see exactly what's on screen, send
keystrokes, wait for commands to finish, and create visual elements —
all programmatically.
Think of it this way: wsh gives you eyes (read the screen), hands (send input), patience (wait for output to settle), and a voice (overlays and panels to communicate with the human).
How It Works
wsh manages terminal sessions via a server daemon and exposes
everything over an HTTP API served on a Unix domain socket. The human
sees their normal terminal. You see a programmatic interface to
the same session. Everything is synchronized — input you send
appears on their screen, output they generate appears in your
API calls. All endpoints are scoped to a session via
/sessions/:name/ prefix (e.g., /sessions/default/input).
By default, wsh serves HTTP over a Unix domain socket at
${XDG_RUNTIME_DIR}/wsh/<name>.http.sock (default name: "default").
This is local-only and requires no authentication. TCP access
is opt-in via --bind for remote scenarios.
MCP also available: wsh is also accessible as an MCP server (14 tools, 3 resources, 9 prompts) via Streamable HTTP at
/mcpon the same socket, or via thewsh mcpstdio bridge. See thewsh:core-mcpprompt for MCP-specific guidance.
Getting Started
Before using the API, make sure a wsh server is reachable.
Step 1: Check for an existing server. A wsh server may already be running — try the default Unix socket:
WSH_SOCK=${XDG_RUNTIME_DIR:-/tmp}/wsh/default.http.sock
curl -sf --unix-socket $WSH_SOCK http://localhost/health
If this returns 200 OK, the server is reachable. Skip to step 3.
Step 2: Start a server (only if needed). Use -L with a unique
name to avoid colliding with any existing wsh instance:
wsh server -L agent-$$ --ephemeral &
sleep 1
WSH_SOCK=${XDG_RUNTIME_DIR:-/tmp}/wsh/agent-$$.http.sock
# Wait for socket to appear
while [ ! -S "$WSH_SOCK" ]; do sleep 0.1; done
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 · 515 lines · 70 tokens per session scan B 4df89da60b46
core is a skill published in the GitHub repository deepgram/wsh (5 stars, last pushed 3mo ago), licensed ISC. It adds 70 tokens to every session and 5,418 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it B with 2 findings (sends data to an external url, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
babysit-pr
Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep…
imagegen
Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…
cpu-profile-analysis
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…
next-cache-components-optimizer
Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…