Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/airiclenz/llama-launchernpx agentmods add skills/airiclenz/llama-launcher/release-llama-launcherWrote 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/airiclenz/llama-launcher/release-llama-launcher)<a href="https://agentmods.dev/skills/airiclenz/llama-launcher/release-llama-launcher"><img src="https://agentmods.dev/badge/skills/airiclenz/llama-launcher/release-llama-launcher/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/airiclenz/llama-launcher/release-llama-launcher"><img src="https://agentmods.dev/badge/skills/airiclenz/llama-launcher/release-llama-launcher.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 | $0.00141 | $0.02389 |
| Opus 5.5 | $0.00056 | $0.00956 |
| Sonnet 5.5 | $0.00028 | $0.00478 |
| Haiku 4.5 | $0.00014 | $0.00239 |
Grade A, and why
release-llama-launcher 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 — 193 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Release llama-launcher — cut and publish a release
Mission
Turn everything shipped since the last published release — the whole history
before the first one — into a published release: notes the owner approved, a
version the owner confirmed, and every channel converged. The deterministic
work belongs to go run ./cmd/release
(ADR-0017: .capcom/knowledge/decisions/adr-0017.md, or
capcom know show adr-0017; the runbook is docs/releasing.md); this skill
does the judgment and the asking.
The optional argument vX.Y.Z (a bare X.Y.Z reads the same) is a version
proposal — the owner still confirms it in step 4. VERSION holds the bare
X.Y.Z; the tag is v + VERSION.
In this repository this skill and release-notes-llama-launcher replace the
personal cut-release and release-notes skills: those stand aside here, and
neither is run for llama-launcher.
Boundaries
- One irreversible step:
publish. Everything before it is local and undone withgit reset --keep HEAD~2; everything after it is fixed forward, never undone. - No bypass. A blocked gate stops the run. You never tag, push, create a
GitHub release, or commit to the tap by hand —
publishdoes all of it, and the tag is made by CI, never by you orpublish. - Ask with the host's user-question tool (
ask_user,AskUserQuestion, or similar) at every Ask below, and wait for the answer. No answer, or a host without the tool, means stop — never assume consent. - Needs an authenticated
gh(gh auth status) and a Go toolchain. Shell snippets assume fail-fast hosts (set -e,pipefail): a command that may match nothing carries|| true— keep it.
1 — Status and resume check (one shell call)
T="$(git rev-parse --show-toplevel)"; cd "$T"
TMP="${TMPDIR:-${TEMP:-/tmp}}"; S="$TMP/release-llama-launcher-status.json"
go run ./cmd/release status --json > "$S" || true
echo "TMP=$TMP"; echo "VERSION=$(cat VERSION)"; git log -2 --format=%s
Read $S yourself (it is small). Its keys: version, highestTag,
gate.{ok,failures}, baseline.{tag,published,range,commits},
steps[].{name,state,detail}. A missing or unparsable file → stop and relay
the command's error (likely gh unauthenticated or no network).
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 · 193 lines · 141 tokens per session scan A dba36697a5ca
release-llama-launcher is a skill published in the GitHub repository airiclenz/llama-launcher (2 stars, last pushed 3d ago), licensed MIT. It adds 141 tokens to every session and 2,389 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 2026-10-14.
Other skills, from other repositories
github
Drive GitHub via the official gh CLI — repos, issues, pull requests, releases, gists, Actions runs, and raw REST through gh api. Use when the user asks to inspect or manage GitHub.
codew-release-qa-sweep
Use before claiming Codewhale release work is done: run the full gate sweep and list the manual QA targets.
release
Skill "release" from Hmbown/Codewhale, covering release, invocation, non-goals and workflow.
apple-calendar
Read macOS Calendar events via the icalBuddy CLI and create events via AppleScript (osascript). Use to check the user's calendar, agenda, upcoming events, or add an event on macOS.
apple-reminders
Manage Apple Reminders via the remindctl CLI on macOS — list, add, complete, delete, manage lists.
obsidian
Read, search, and create Markdown notes inside an Obsidian vault on disk.