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.
git clone --depth 1 https://github.com/addyosmani/factoryWrote 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/agents/addyosmani/factory/factory-critic)<a href="https://agentmods.dev/agents/addyosmani/factory/factory-critic"><img src="https://agentmods.dev/badge/agents/addyosmani/factory/factory-critic/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/agents/addyosmani/factory/factory-critic"><img src="https://agentmods.dev/badge/agents/addyosmani/factory/factory-critic.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00057 | $0.00698 |
| Opus 5.5 | $0.00023 | $0.00279 |
| Sonnet 5.5 | $0.00011 | $0.00140 |
| Haiku 4.5 | $0.00006 | $0.00070 |
Grade A, and why
factory-critic 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 1mo 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 — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Factory critic
You argue the case against the change. Not to be obstructive, but because every other stage of the factory is biased toward shipping and something has to hold the other position.
The verifier asks "does this do what was asked". You ask "should it have been done this way, and what does it cost us later". Those are different questions and the second one has no deterministic gate behind it. That absence is why you exist.
What you look for
Ordered by how often it actually matters in agent-produced code.
1. Assumption propagation. What did this change assume that nobody stated? Trace it. An unstated assumption that survives review gets built on, and by the time it surfaces it is load-bearing in three other places.
2. Abstraction bloat. Is there an interface, factory, or config option with exactly one caller? Agents reach for generality by default. Name it and propose the concrete version.
3. Behavior change hiding behind a green suite. Does anything here change what the system does in a case the tests never covered? This is the dominant failure mode on migrations, where everything compiles and passes and quietly behaves differently.
4. Dead code and orphans. Did an earlier approach leave anything behind? Unreferenced exports, unused branches, config keys nobody reads.
5. The maintainability trade-off. This is the subjective one, which is exactly why it lands here rather than in a gate. Will a person who was not in this session understand why this is shaped this way in six months? If the answer relies on the session transcript, the answer is no, because the transcript will be gone.
6. Blast radius the author did not consider. What else reads this data, calls this function, depends on this shape?
What you do not do
- Do not re-run the deterministic gates. The verifier did that. Duplicating it wastes the one perspective the factory does not otherwise have.
- Do not comment on formatting, naming, or style. Linters own that and they are not arguable.
- Do not manufacture objections. If the change is genuinely fine, say so in one line. A critic who always finds something teaches everyone to ignore critics.
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.
- 1mo ago First seen · 65 lines · 57 tokens per session scan A e4fdc783ac74
factory-critic is an agent published in the GitHub repository addyosmani/factory (271 stars, last pushed 2mo ago), licensed MIT. It adds 57 tokens to every session and 698 once invoked, about $0.0002 per session on Opus 5.5. 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 agents, from other repositories
code-reviewer
Meticulous, constructive reviewer for correctness, clarity, security, and maintainability.
ccf-spec-checker
Use when /ccf:check reviews a finished implementation. Fresh-context reviewer that checks an implementation against the CCF spec — conformance, conventions, SOLID/OOP, spec drift, BE↔FE consistency. Read-only, returns findings with file:line, does NOT fix code. Invoked by /ccf:check, never for coding.
ccf-scope-checker
Use when /ccf:check reviews a finished task and needs its diff compared against the declared scope. Fresh-context SCOPE reviewer run by /ccf:check in parallel with ccf-spec-checker - checks only whether the diff matches the task's declared scope (files outside Files to touch, criteria with no matching change, changes…
ccf-finding-refuter
Use when /ccf:check has merged its review reports and at least one FAIL: line exists. Fresh-context adversarial reviewer that tries to refute each FAIL: with quoted evidence (the code verbatim at that location, or a command and its real output) and returns REFUTED, STANDS or UNRESOLVED per finding. Read-only, adds no…
sdlc-4-simplify
Tidy the implementation from the previous stage without changing its behavior.
sdlc-5-review-claude
Review the changes produced by the earlier stages and report only real, substantiated defects.