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 commands/melagiri/code-insights/start-reviewgit clone --depth 1 https://github.com/melagiri/code-insightsWrote 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/commands/melagiri/code-insights/start-review)<a href="https://agentmods.dev/commands/melagiri/code-insights/start-review"><img src="https://agentmods.dev/badge/commands/melagiri/code-insights/start-review.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 | $0.00000 | $0.04004 |
| Opus 5 | $0.00000 | $0.02002 |
| Sonnet 5 | $0.00000 | $0.00801 |
| Haiku 4.5 | $0.00000 | $0.00400 |
Grade A, and why
start-review 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 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- **For API changes**: curl output showing the endpoint returns the expected response shape How it starts
The opening of the file, as written. The whole thing — 376 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/start-review — Triple-Layer Code Review Team
PR: $ARGUMENTS
You are setting up a triple-layer code review for PR $ARGUMENTS. This can be used standalone or as part of a /start-feature team workflow. The review loops until all FIX NOW items are resolved — it does NOT end after a single pass.
Step 1: Get PR Details
Fetch the PR details:
# Get the correct owner from git remote
git remote get-url origin | sed 's/.*[:/]\([^/]*\)\/[^/]*\.git/\1/'
Use gh pr view $ARGUMENTS to get PR title, description, and diff stats.
Use gh pr diff $ARGUMENTS to get the diff.
Determine the PR scope:
- Count files changed and lines changed per file
- Identify if it touches schema concerns (types.ts, SQLite schema, server API)
- Classify changed files into domains for specialist selection (Step 2)
Step 2: Select Domain Specialists
Instead of fixed generalist reviewers, select 1-2 domain specialists based on what the PR actually changes. Each specialist combines deep domain expertise with general engineering awareness. See docs/REVIEW-SPECIALISTS.md for the full specialist registry and prompt templates.
Domain Classification
Classify each changed file into a domain based on file patterns:
| Domain | File Pattern Triggers |
|---|---|
| SQL/Database | server/src/routes/*.ts (when containing SQL keywords), cli/src/db/*.ts |
| React/Frontend | dashboard/src/**/*.tsx, dashboard/src/**/*.ts |
| Node/CLI | cli/src/commands/*.ts, cli/src/utils/*.ts, server/src/*.ts (non-route, non-LLM) |
| Parser/Provider | cli/src/providers/**/*.ts, cli/src/parser/*.ts |
| LLM | server/src/llm/**/*.ts — handled by LLM Expert (separate, unchanged) |
Files that don't match any domain (e.g., package.json, tsconfig.json, .gitignore, CLAUDE.md) are ignored during classification.
Selection Algorithm
- For each domain, sum the lines changed across all files in that domain
- Primary specialist = domain with the most lines changed
- Secondary specialist = next domain IF it has >= 30% of total changed lines AND is a different domain
- Cap at 2 domain specialists (LLM Expert is a separate conditional reviewer)
- If a PR changes ONLY non-domain files, use the Node/CLI Specialist as general-purpose
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 First seen · 376 lines · 0 tokens per session scan A ed85be5bee17
start-review is a command published in the GitHub repository melagiri/code-insights (76 stars, last pushed 3mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 4,004 tokens. 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 commands, from other repositories
ship-and-babysit
Commit, push to origin (fork), open PR to tinyhumansai/openhuman:main, then poll every 5min for CodeRabbit comments and CI failures, resolve them, and exit when clean.
rp-build-cli
Build with rp-cli context builder → chat → implement.
lians-recall
Recall current (non-stale) facts from Lians memory, optionally as-of a past date.
netllm-connect
Wire Cursor, Claude Code, Codex, or Honcho to the local netllm router.
netllm-setup
First-time netllm install from this repo (uv sync, init, discover, verify).
netllm-doctor
Troubleshoot netllm misconfigurations (PATH, providers, mDNS, agent health).