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/devstride/claude-plugin/setupnpx skills add devstride/claude-plugin --skill setupgit clone --depth 1 https://github.com/devstride/claude-pluginWhat 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 | $0.00077 | $0.07955 |
| Opus 5 | $0.00039 | $0.03978 |
| Sonnet 5 | $0.00015 | $0.01591 |
| Haiku 4.5 | $0.00008 | $0.00796 |
Grade A, and why
setup 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 — 477 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Human output. Read ${CLAUDE_PLUGIN_ROOT}/skills/build-item/references/plain-language-output.md once per top-level run; composed skills reuse it. Apply it to every message.
Set the delivery loop up for this repository and DevStride organization: inspect (A), report (B), map work types onto roles (C), ask what is left (D), write (E) — or merge into an existing config (F) — then prove it by running it (G).
Optional argument: $ARGUMENTS
validate— skip to Phase G on the config already there; nothing asked or written. Run A1 first even so — Phase G assumes the repository root.- A detector name —
ecosystem(A2–A3),verify(A4),ci-inspect(A5),branches(A6),engines(A7) — inspect and report just that part. (ci-inspect, notci: the bare word is the write mode; one token means one thing.) docs— the documentation-hooks mode, the one narrowed run that WRITES: A1 + A8, Phase D's three documentation questions plusrelease.deployVerification, then write only thedocsblock and that key (create-with-just-those-keys or Phase F merge,release.docsRepomigration included), scaffold E2, run Phase G check 7 alone.ci— the CI-cost mode, the second narrowed run that WRITES, and the only mode that touches a workflow file: A1 + A5; detect which of the four mechanics in${CLAUDE_PLUGIN_ROOT}/skills/setup/references/ci-cost-patterns.mdeach pull-request workflow carries; show the exact diff per mechanic missing OR present-but-inert; apply only accepted diffs — nothing else rewritten. Writeci.freezeBaseWhileReleasePrReadyandci.expectedRunsPerPullRequestif absent. Draft-gate diffs accepted while the threereview.*CI-ordering booleans arefalse→ propose flipping themtruein the same change. Run Phase G's CI checks. Offer/devstride:ci-auditfirst for numbers.- Nothing — the full run: inspect, ask, write, scaffold, validate.
Run each mode's prerequisites silently — A1 always, A2–A3 before A4. A narrowed run reports
fewer keys, never worse ones, stops after Phase B and writes nothing (a write would fill
every unexamined key with a default); a full /devstride:setup is what writes. docs and
ci are the exceptions by design — each writes exactly the keys its own questions answer.
The write boundary — the whole of it: .claude/ds-config.json; in E2 only, the scaffolded
skills (never overwriting); in ci mode only, the pull-request workflows under
ci.workflowGlobs (each change a diff, applied on an explicit yes). Nothing else, ever — no git
mutation, no installs, no DevStride write tool. Phases A–D are strictly read-only:
git status --porcelain stays byte-identical until the write.
Why detection comes before questions
Detect before you ask — and never guess: a wrong detected value arrives wearing the authority of evidence. Three outcomes; the third is not a failure:
| Status | Meaning | Later |
|---|---|---|
detected |
One unambiguous answer, with evidence | A prefilled default |
ambiguous |
Several plausible answers, or one needing confirmation | A question offering the candidates |
unknown |
The repository does not answer this | A question with no prefill |
Every row carries the evidence that produced it — a row without evidence is a guess with a status attached.
What ships with it
10 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.
- references/ci-cost-patterns.md 13 KB
- references/config-defaults.md 23 KB
- references/detector-evidence.md 11 KB
- references/docs-skill-templates/update-documentation.md 4.1 KB
- references/docs-skill-templates/update-release-notes.md 3.4 KB
- references/interview-rationale.md 4.2 KB
- references/review-engines.md 4.8 KB
- references/statusline-segments.md 5.4 KB
- references/statusline.sh 11 KB runs code
- references/validation-checklist.md 14 KB
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 · 477 lines · 77 tokens per session scan A 87a5d433ea5f
setup is a skill published in the GitHub repository devstride/claude-plugin (2 stars, last pushed 3d ago), licensed MIT. It adds 77 tokens to every session and 7,955 once invoked, about $0.0004 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
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…