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 oliver-kriska/claude-elixir-phoenix --skill watch-prgit clone --depth 1 https://github.com/oliver-kriska/claude-elixir-phoenixWrote 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/oliver-kriska/claude-elixir-phoenix/watch-pr)<a href="https://agentmods.dev/skills/oliver-kriska/claude-elixir-phoenix/watch-pr"><img src="https://agentmods.dev/badge/skills/oliver-kriska/claude-elixir-phoenix/watch-pr/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/oliver-kriska/claude-elixir-phoenix/watch-pr"><img src="https://agentmods.dev/badge/skills/oliver-kriska/claude-elixir-phoenix/watch-pr.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00041 | $0.00949 |
| Opus 5 | $0.00020 | $0.00475 |
| Sonnet 5 | $0.00008 | $0.00190 |
| Haiku 4.5 | $0.00004 | $0.00095 |
Grade A, and why
watch-pr 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 10d 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- phx-watch-pr — 89% identical, 7 lines differ
How it starts
The opening of the file, as written. The whole thing — 94 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Watch PR — Amp-Native Lifecycle
Use the elixir_phoenix_watch_pr plugin tool. Do not launch gh pr checks --watch, gh run watch, a foreground sleep loop, or the copied Claude watcher
script: none of those owns an Amp Orb keep-alive lease.
Iron Laws
- Acquire one bounded keep-alive lease — start the plugin lifecycle once, then let its durable state and dedupe handle polling and plugin reloads.
- Deployment is never readiness — deployment, release, preview,
production, and prod checks are reported under
excludedChecks; they are never counted as passing or failing required CI. - Never merge or deploy — watching and
--fixauthorize neither action. --fixuses the same worker thread — the plugin serializes one concise event turn at a time and directs that turn to the installedphx-pr-reviewworkflow for actionable review feedback or failed required CI. Validate the cause before editing.- Every watch is bounded — timeout and every other terminal path release the lease. A timeout is incomplete, never success.
Start
Parse the PR number/URL and flags, then call:
elixir_phoenix_watch_pr {
"action": "start",
"pr": "<number-or-url>",
"checksOnly": false,
"fix": false
}
Map --checks-only and --fix to booleans. Map --max-hours,
--quiet-minutes, and --poll-seconds to maxDurationHours,
quietPeriodMinutes, and pollIntervalSeconds. Defaults are 2 hours maximum,
15 minutes quiet, and 60 seconds polling. Report the tool's initial snapshot
and then stop taking turns; the plugin wakes this thread only for actionable
failures or feedback and terminal outcomes.
Event Handling
- Unchanged snapshots do not wake the model.
- A changed head SHA resets readiness but does not itself wake the model.
- Failed or cancelled required non-deployment CI wakes the model once per distinct actionable snapshot. Pending and passing progress keeps polling inference-silent.
- Excluded deployment-like checks appear separately in explicit status and terminal summaries. Their transitions do not reset readiness or wake the model.
- Any unresolved review thread keeps the lease active.
- A head push, required-check transition, review, or comment resets the quiet clock. This includes a review with no unresolved thread or dedicated check, but routine non-actionable activity does not wake the model.
- With
--fix, unresolved feedback explicitly loadsphx-pr-review; required CI failures include check names and links. Re-fetch threads or inspect logs, fix only valid branch-owned causes, verify, push the authorized PR branch update, reply, and resolve where appropriate. Do not blindly rerun shared CI. - Required CI must remain green with zero unresolved threads and no relevant activity for the full quiet period before success releases the lease. Never accept an early green snapshot as immediate readiness.
What ships with it
1 file 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.
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.
- 10d ago First seen · 94 lines · 41 tokens per session scan A b7b4cb3261c7
watch-pr is a skill published in the GitHub repository oliver-kriska/claude-elixir-phoenix (542 stars, last pushed 4d ago), licensed MIT. It adds 41 tokens to every session and 949 once invoked, about $0.0002 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
craft-plugin-release
Releasing Craft CMS plugins — tagging, Packagist propagation, GitHub releases, branch promotion, shared-library ordering, history rewrites. ALWAYS load when cutting, preparing, verifying, or debugging a plugin release: bumping a version, dating a changelog, creating or moving a git tag, or checking what Packagist…
git-for-research-code
When the user wants to version-control optimization research code - small commits per experiment change, tags for paper result snapshots, .gitignore for solver logs, linking result tables to commit hashes, and branch strategy for risky refactors. Also use when the user mentions "git workflow," "version control…
git-ops
Use when Sam asks Claude Code to create, rename, compare, commit, push, merge, promote, or clean up Git branches. Covers Sam's feature branch naming convention, safe Git pre-checks, JetBrains DontCommit handling, commit-message defaults, push and release promotion verification, branch rename upstream cleanup, branch…
skill-pr-review
Comprehensive code review for pull requests using parallel multi-agent analysis. Audits CLAUDE.md compliance, checks for bugs, analyzes git history, reviews comments, and filters by confidence score. Use when reviewing a GitHub PR, mentions "code review", "review this PR", or "/code-review".
setup-pre-commit
Set up Husky pre-commit hooks with lint-staged (Prettier), type checking, and tests in the current repo. Use when user wants to add pre-commit hooks, set up Husky, configure lint-staged, or add commit-time formatting/typechecking/testing.
odoo-git-commit
Use when committing Odoo work — before git commit, when a staged diff spans multiple modules or mixes a move with edits, or when writing a commit message or PR body. Covers splitting into one-logical-change commits, the [TAG] module: summary convention, move-then-modify, and terse human messages over AI essays. Invoke…