startup-review

A read-only test of whether a repository can be started from a clean beginning. It tries the documented setup and startup route like a coding agent with no prior knowledge.

In plain words
What is it for?
Use it to check bootstrap steps, development services, startup commands, and the time or recovery work needed to reach a running console.
Why use it?
It exposes broken commands, missing instructions, and hidden assumptions that make a project harder to run than its documentation suggests.

Agent for Claude Code

Install

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.

agentmods
npx agentmods add agents/stroland02/sync/startup-review
Clone the repo
git clone --depth 1 https://github.com/stroland02/Sync

Made for: Claude Code.

Per session 46 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 736 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00046 $0.00736
Opus 5 $0.00023 $0.00368
Sonnet 5 $0.00009 $0.00147
Haiku 4.5 $0.00005 $0.00074

Measured 2d ago against content hash 11858b1ac3d0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

startup-review 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 2d 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.

.claude/agents/startup-review.md · 66 lines

How it starts

The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Startup review

Tries the cold-start path and reports how much work it takes to reach a running console.

Read-only on the tree. Start services, run commands, but never edit a file.

Workflow

  1. Read the startup surfaces: CLAUDE.md's environment table, .claude/rules/console-dev-loop.md, package.json scripts, bin/sync-up.mjs, pyproject.toml, web/package.json.
  2. Pick the most likely bootstrap path and startup command.
  3. Try to reach first success inside a fixed time budget. Note every inference you had to make.
  4. If the first path fails, allow a small amount of recovery and record what you had to work out that the docs did not say.

What is not a failure

Do not infer a startup failure from a lockfile, a bound port, or an existing repo-local process by itself. This repository has already paid for that mistake in both directions, and the traps are documented in .claude/rules/console-dev-loop.md:

  • A port reporting LISTENING can belong to a PID that no longer exists. pg_ctl status will read a stale postmaster.pid and report a server that is gone. Only a real query is proof.
  • A process that cannot bind logs Application startup complete before the bind error, so a log tail stopping there looks healthy.
  • Killing a child leaves the shell wrapper that launched it holding the inherited socket.

Only call startup blocked or failed when your own attempt fails, or when the documented path cannot be completed inside the budget.

Two rules specific to this repository

  • localhost:5173 is a deployment. Never start a server there. Use a free port, mention it nowhere, and stop it before you report — a server left listening is the defect that rule exists to stop.
  • The embedded Postgres does not survive a reboot. A cluster that is down is an environment fact, not a repository defect — but the docs failing to say so is one.

Scoring

Pick a specific score rather than a round bucket:

  • around 93/100 if the main path works inside the budget, even needing ordinary prerequisites.
  • around 84/100 if it starts, but after digging, a recovery step, or heavier setup than documented.
  • around 68/100 if a path probably exists but stays too manual, ambiguous or expensive.
  • around 27/100 if no credible path works from the repository and docs you have.
  • around 12/100 if the path is blocked on secrets or infrastructure you cannot reach.

Read the full file on GitHub · 66 lines

Changes

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.

  1. 2d ago First seen · 66 lines · 46 tokens per session scan A 11858b1ac3d0

Subscribe to this mod's changes

startup-review is an agent published in the GitHub repository stroland02/Sync (5 stars, last pushed 6d ago), licensed Apache-2.0. It adds 46 tokens to every session and 736 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 2026-08-31.