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/blakemartz/delivery-loop/initnpx skills add blakemartz/delivery-loop --skill initgit clone --depth 1 https://github.com/blakemartz/delivery-loopWhat 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 | $0.00057 | $0.00850 |
| Opus 5 | $0.00028 | $0.00425 |
| Sonnet 5 | $0.00011 | $0.00170 |
| Haiku 4.5 | $0.00006 | $0.00085 |
Grade A, and why
init scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 39 lines — stays where its author put it; the contents beside it link to each section on GitHub.
init — bootstrap this repo for the delivery loop
One-time, idempotent setup. Runs the bundled bootstrap, then walks the human through the judgment calls the script can't make.
Protocol
-
Run:
bash "${CLAUDE_PLUGIN_ROOT}/scripts/delivery-init.sh" $ARGUMENTS(any arguments become extramodule:<name>labels). Local scaffolding runs first —.claude/delivery.conf,.gitignore, a seededscripts/check.sh(the gateGATE_CMDpoints at), and aSTYLEGUIDES_DIRstub — then the GitHub steps (labels). -
No GitHub repo connected (script exits 2). The backlog lives in GitHub Issues, so this is a hard prerequisite. Ask the human which they want (outward-facing — never do it unasked):
- a brand-new repo:
gh repo create <name> --private --source=. --push(confirm name and visibility first), or - connect an existing one:
git remote add origin <url>and push the base branch. Then re-run the script — it's idempotent and will finish the labels.
- a brand-new repo:
-
Issues disabled (script warns). Common on forks. Ask the human to confirm, then run
gh repo edit --enable-issues— without Issues the loop is inert. -
Establish the gate — the thing that needs the most judgment. Open the seeded
scripts/check.shand reconcile it with what the repo actually runs, preferring the repo's own declared commands over guessed ones. Checkpackage.jsonscripts, aMakefile/justfilecheck/testtarget, and CI steps in.github/workflows/for the real commands.- Tooling detected: show the human the seeded
check.sh, note any declared command it should use instead (e.g. the repo has amake checkthat bundles more than init found), and adjust on their confirmation. - No tooling detected (the seed exits 1 by design — an undefined gate is a failing gate, never a green one): tell the human plainly that no linter/type-checker/test framework was found, and suggest a sensible starter set for the detected language (e.g. ruff + mypy + pytest for Python; eslint + tsc + vitest for TS;
go vet/build/testfor Go). Wire in whatever they approve; leave the failing guard if they defer.
- Tooling detected: show the human the seeded
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.
- yesterday First seen · 39 lines · 57 tokens per session scan A aaee38be0359
init is a skill published in the GitHub repository blakemartz/delivery-loop (4 stars, last pushed 1mo ago), licensed MIT. It adds 57 tokens to every session and 850 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 2026-08-31.
Other skills, from other repositories
zh-output-example
Reviews Chinese skill documentation for publication readiness. Invoke when the user asks in Chinese to review a skill.
peekaboo
Capture and automate macOS UI with the Peekaboo CLI.
powerpoint
Create designed, editable PowerPoint .pptx presentations with PptxGenJS. Use when the user asks to create, generate, update, or inspect a deck, slide deck, presentation, or .pptx file.
oracle
Best practices for using the oracle CLI (prompt + file bundling, engines, sessions, and file attachment patterns).
argent-tv-interact
Control and inspect TV apps via argent — Apple TV (tvOS), Android TV (leanback), and Amazon Fire TV (Vega). Boot the target, read focus, navigate with the D-pad remote, type, screenshot, and on Vega debug the JS runtime (evaluate, console logs, network inspector). Use when a task targets a TV (runtimeKind "tv", or…
review-offered-task
Review a task that has been offered to you and decide whether to accept or reject it.