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/comet-ml/opik/explore-featurenpx skills add comet-ml/opik --skill explore-featuregit clone --depth 1 https://github.com/comet-ml/opikWhat 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 | $0.00044 | $0.03519 |
| Opus 5 | $0.00022 | $0.01759 |
| Sonnet 5 | $0.00009 | $0.00704 |
| Haiku 4.5 | $0.00004 | $0.00352 |
Grade A, and why
explore-feature scanned grade A with 1 finding 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
build, not just that a FE answers**: for a FE-only PR merged to main, `curl How it starts
The opening of the file, as written. The whole thing — 210 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Explore Feature
Turns "here's my change, cover it" into one committed, locally-green Playwright spec.
It is a thin orchestrator: it owns two phases — resolve what to cover, and confirm the result —
and delegates the actual authoring (analyze FE, discover live UI, write POM/spec, run green) to
the writing-e2e-tests skill.
Announce at start: "I'm using the explore-feature skill to add an e2e test for X."
What this does — and doesn't
- Does: resolve the change surface → pick the one flow worth covering → hand it to
writing-e2e-tests→ confirm the spec is tagged, in the taxonomy, and green. - Doesn't: deep bug-hunting or edge-case coverage (that's a separate QA activity); CI wiring.
- Cheap per-PR happy-path only. One flow, green locally in minutes.
The output is an ordinary spec: tests/<area>/<name>.spec.ts, a tier tag, an @area:, a @cap:
per test — exactly what every other spec in the estate looks like, and picked up by the same
suites. There is no separate lane and no version stamp.
The loop
digraph explore_feature {
rankdir=TB;
"1. Resolve scope (GATE)" [shape=box];
"2. Local-run gate (GATE)" [shape=box];
"3. Delegate authoring to writing-e2e-tests" [shape=box];
"4. Confirm tagged + green" [shape=box];
"1. Resolve scope (GATE)" -> "2. Local-run gate (GATE)";
"2. Local-run gate (GATE)" -> "3. Delegate authoring to writing-e2e-tests";
"3. Delegate authoring to writing-e2e-tests" -> "4. Confirm tagged + green";
}
Phase 1 — Resolve scope (gate)
Normalize whatever the dev pointed at into one ScopeSpec before authoring anything.
Input modes (auto-detect from the argument; ask if ambiguous):
| Mode | Trigger | Resolve |
|---|---|---|
| Local diff | no arg / dirty tree / "my branch" / "my changes" | see Local-diff mode below — this is the default when no PR is named, incl. "I haven't opened a PR yet" |
| A PR | #<n> or a PR URL |
PR diff + linked ticket via gh pr view <n> --json … (or the GitHub MCP) |
| Multi-PR / multi-ticket | a list | union of the diffs |
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 · 210 lines · 44 tokens per session scan A c614d79aaa01
explore-feature is a skill published in the GitHub repository comet-ml/opik (21,685 stars, last pushed yesterday), licensed Apache-2.0. It adds 44 tokens to every session and 3,519 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
amazon
Shop on Amazon and Amazon Fresh through your browser.
html-deck-builder
Plan and build browser-editable HTML presentations with narrative contracts, responsive slide layouts, keyboard navigation, offline-safe assets, and render verification. Use when the final presentation should be an HTML file or browser-based deck rather than a native PPTX.
magazine-web-deck
Create editorial or Swiss-inspired single-file web presentations with strong typographic hierarchy, horizontal navigation, responsive composition, and restrained motion. Use for keynote-like web decks, editorial stories, launches, and visually distinctive browser presentations.
textual-screenshot
Capture a Textual terminal UI as an SVG using its headless test harness. Use when asked to make, attach, or preview a screenshot of deepagents-code/dcode or another Textual app, visually verify a TUI state, or render a modal, screen, or widget without a desktop or browser.
playwright
Use when the task requires capturing or automating a real browser from the terminal.
search-params
URL search param and hash state management. Use when adding or modifying URL search params, working with useSearchParams, setSearchParams, useSearchParamState, or navigate() with query strings or hash fragments, or fixing browser back/forward button issues.