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/jackfranklin/dotfiles/review-plannpx skills add jackfranklin/dotfiles --skill review-plangit clone --depth 1 https://github.com/jackfranklin/dotfilesWrote 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/jackfranklin/dotfiles/review-plan)<a href="https://agentmods.dev/skills/jackfranklin/dotfiles/review-plan"><img src="https://agentmods.dev/badge/skills/jackfranklin/dotfiles/review-plan.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00049 | $0.01451 |
| Opus 5 | $0.00024 | $0.00726 |
| Sonnet 5 | $0.00010 | $0.00290 |
| Haiku 4.5 | $0.00005 | $0.00145 |
Grade A, and why
review-plan 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 5d 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 — 79 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are presenting a plan for human review using the ai-review CLI, in interactive mode: the CLI stays running across multiple rounds, and you revise the plan file in place instead of restarting the CLI for each pass.
Steps
-
Write the plan to a file in the
/tmp/directory (e.g./tmp/plan-<timestamp>.md). -
Generate AI annotations. Before opening the review, write a JSON file with a summary and any per-line notes to guide the reviewer. This is especially useful on later rounds to show what changed since the last one.
Write the file to
/tmp/annotations-<timestamp>.jsonusing this schema:{ "summary": "One or two sentences: what this plan does, or what changed since the last round.", "annotations": [ { "startLine": 15, "endLine": 22, "text": "This section was rewritten to address the feedback about error handling." } ] }Rules for generating annotations:
summaryis optional but strongly recommended; always write one after the first round.annotationsis optional; include only lines worth drawing the reviewer's attention to.- Do not include a
filefield — plan mode uses plain line numbers only. startLineandendLineare 1-indexed line numbers in the plan file. To get accurate numbers: read the written plan file back with line numbers (e.g.cat -n /tmp/plan-<timestamp>.md), then reference the specific lines.- Fenced code blocks and tables are treated as a single block. Annotating any line inside a code fence attaches the annotation to the opening
```line. If you want to annotate content within a fence, use the line number of the opening fence. - Read the written annotations file back and verify line numbers look correct before proceeding. If annotations don't appear in the review UI, they were silently dropped with no error — check that line numbers fall within the rendered content.
-
Start the CLI in the background with
--interactive, so you can keep working while it stays open across rounds. Choose a title that is short (3–6 words) and specific to the current task — the user may have multiple review tabs open at once and needs to tell them apart at a glance:node ~/git/ai-review-plan/dist/cli.js plan \ --title "<short task-specific title>" \ --theme <dark|light> \ --ai-annotations-file /tmp/annotations-<timestamp>.json \ --interactive \ /tmp/plan-<timestamp>.mdUse
--theme lightunless the user has expressed a preference for dark mode. -
Wait for stdout to print
Watching: <path>— this confirms the server is up and the browser has been asked to open. The process keeps running after this; do not wait for it to exit. -
Keep monitoring the background process's stdout by polling:
- Schedule a 10-second timer using the
scheduletool (e.g.,DurationSeconds="10",TimerCondition="never"). - When the timer fires, check the background task's status using
manage_taskwith actionstatus, read its log file usingview_file, and inspect the latest output. - If no new output or feedback is present, immediately schedule another 10-second timer to continue polling.
- Each round ends in one of three ways:
=== FEEDBACK END ===/[ACTION REQUIRED]— the user clicked Request Changes. The stdout since the last marker contains the## Review: CHANGES REQUESTEDcomments. You must act immediately and autonomously. Do not wait for a user chat prompt or command termination. Read the comments carefully, revise the plan file in place at its original path (do not write a new file — the CLI is watching that exact path), and optionally rewrite the annotations file with an updated summary describing what you just changed. Once you save the file, the changes are automatically sent to the browser. Then go back to waiting on stdout.- The process exits with code 0 — the user clicked Approve. Stdout will contain
## Review: APPROVEDand any final comments; check for inline comments and address them, then proceed. This ends the review — go to step 6. === SESSION CLOSED: client disconnected ===(process exits non-zero) — the browser tab didn't reconnect within 30 seconds (e.g. the user closed it without submitting). Tell the user the session closed and ask whether to re-open it (re-run from step 3 against the same plan file) or stop here.
- Schedule a 10-second timer using the
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.
- 5d ago First seen · 79 lines · 49 tokens per session scan A f5cc0c53b1a0
review-plan is a skill published in the GitHub repository jackfranklin/dotfiles (254 stars, last pushed 4d ago), licensed MIT. It adds 49 tokens to every session and 1,451 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
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
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…
chat-pet-sprite-creation
Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.
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…
insight-error-page
Write or audit an insight-kind error page for the Next.js dev overlay. Use when creating a new errors/ .mdx page, auditing an existing one, or checking that a page matches the framework fix cards. Covers page structure, title alignment, FixCard cards with Copy prompt button, code snippets, terminology verification…