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 cosmix/loom --skill loom-searchgit clone --depth 1 https://github.com/cosmix/loomWrote 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/cosmix/loom/loom-search)<a href="https://agentmods.dev/skills/cosmix/loom/loom-search"><img src="https://agentmods.dev/badge/skills/cosmix/loom/loom-search.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00011 | $0.03900 |
| Opus 5 | $0.00005 | $0.01950 |
| Sonnet 5 | $0.00002 | $0.00780 |
| Haiku 4.5 | $0.00001 | $0.00390 |
Grade A, and why
loom-search 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 4d 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 — 263 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Search
Overview
Full-text search on Lucene-based engines (Elasticsearch/OpenSearch) plus the leaner alternatives (Meilisearch/Typesense). The engine is easy to stand up and easy to get subtly wrong: analyzer mismatches, facet counts that fight their own filters, deep pagination that falls over at 10k, and relevance that looks fine on your three test queries and terrible in production. This skill targets those.
text (analyzed, full-text, scored) vs keyword (exact, aggregatable, sortable, filterable) is the decision under most of these. Get the mapping right first.
Analysis: the root of most bugs
An analyzer = optional char filters → one tokenizer → token filters. It runs at index time (on the stored field) and at query time (on the search string). The inverted index only ever contains analyzed tokens.
⚠ Index-time / query-time analyzer mismatch is the #1 silent search bug. If you index with an edge_ngram analyzer and also analyze the query with it, searching "cat" expands to c, ca, cat and matches "category", "catalog", "cathedral" — garbage relevance. The fix is almost always: aggressive analyzer at index time, plain analyzer at search time.
"name": {
"type": "text",
"analyzer": "autocomplete", // edge_ngram — index time only
"search_analyzer": "autocomplete_search" // just lowercase — query time
}
- Reindex required to change the index-time analyzer (existing tokens are already committed).
search_analyzercan change without reindex. - normalizer = the
keywordequivalent of an analyzer (lowercase/asciifold only, no tokenizer) so exact-match/aggregation fields can be case-insensitive. - Verify what a field actually produces with
GET /index/_analyze— don't guess.
POST /products/_analyze
{ "field": "name", "text": "Wireless Headphones" } // shows the exact tokens indexed
Common token filters: lowercase, asciifolding (café→cafe), stop (drop the/a/is — omit for short-field/name search), stemmer/snowball (running→run), synonym/synonym_graph.
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.
- 4d ago Changed · -34 tokens per session 0e73e5c8ab43
- 8d ago First seen · 263 lines · 45 tokens per session scan A c872580352cd
loom-search is a skill published in the GitHub repository cosmix/loom (54 stars, last pushed yesterday), licensed MIT. It adds 11 tokens to every session and 3,900 once invoked, about $0.0001 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 2026-08-30.
Other skills, from other repositories
contentful-migration
Write and run Contentful content model migration scripts using the contentful-migration library and the Contentful CLI. Covers creating, editing, and deleting content types and fields, validations, editor interface configuration, editor layouts, sidebar widgets, entry transformations, tags, annotations, and the…
apple-cktool
Operates Apple's Xcode-bundled cktool command-line utility for CloudKit development automation on macOS. Use when Codex needs to run, script, explain, or troubleshoot xcrun cktool; manage .ckdb schemas with export, validate, import, or reset; discover CloudKit teams; create, query, or delete test records; configure…
API Discoverability for Agents
Making self-hosted services agent-discoverable — bake in a machine-readable API description (OpenAPI spec or a minimal API.md) when building, and discover-first (spec paths, repo search) before probing when integrating.
supabase
Configure Supabase for Next.js SaaS: Drizzle + pooler, Google-only Auth, RLS, cost optimization. Use when setting up Supabase, connecting Drizzle, SSR auth, or optimizing costs.
pw-module-fieldtype-inputfield
Use when building custom database fieldtypes and their corresponding interface inputfields in ProcessWire.
cqrs-implementation
Implement Command Query Responsibility Segregation for scalable architectures. Use when separating read and write models, optimizing query performance, or building event-sourced systems.