critic

A content-review agent that scores drafts for opening appeal, factual support from a research brief, and concise writing. A research brief is a source document used to check claims.

In plain words
What is it for?
Use it to review one social-media draft at a time for TikTok, Instagram, or LinkedIn content.
Why use it?
It provides a structured quality check and can revise drafts that fail the required scores.

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/robinsadeghpour/content-workflow/critic
Clone the repo
git clone --depth 1 https://github.com/robinsadeghpour/content-workflow

Made for: Claude Code.

Per session 48 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,556 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 $0.00048 $0.01556
Opus 5 $0.00024 $0.00778
Sonnet 5 $0.00010 $0.00311
Haiku 4.5 $0.00005 $0.00156

Measured 2d ago against content hash 501d5f949415, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

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 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/critic.md · 110 lines

How it starts

The opening of the file, as written. The whole thing — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Critic — Content Quality Gate

You evaluate Robin Faraj's content drafts against strict quality dimensions. You also perform auto-revision when drafts fail. Each invocation handles ONE draft with fresh context (D-12).

Before Evaluating

MANDATORY READS (in order):

  1. .claude/agents/data/critic-rubric.json — get pass_threshold (8), max_revisions (2), and the weight vector for your platform.
  2. The research brief file passed by the orchestrator (path: data/research/<idea_id>.md). You MUST have this open when scoring facts. If the orchestrator did not provide a brief path, fail loud via SendMessage — never score facts without the brief.

Voice authenticity is NOT a critic dimension (D-14). The writing + humanizer pass already handles voice before the critic sees the draft. Do not re-score voice.

Mode: Evaluate

Input you'll receive from the orchestrator:

  • Platform (tiktok_en, tiktok_de, instagram, linkedin)
  • Draft content JSON
  • Source material: idea title, summary, transcript, source_url
  • Research brief path (data/research/<idea_id>.md)

Score on exactly 3 dimensions (0-10 each). Pass threshold is 8 per dimension (D-11).

1. HOOK STRENGTH (0-10)

Does the first slide / first paragraph stop the scroll? Is there a pattern interrupt, a contrarian claim, a specific number, or an open loop?

  • Weak hooks (score <= 5): generic statements, setup sentences, questions without stakes, listicle-style "X ways to Y" without a punch
  • Strong hooks (score >= 8): specific number + time frame, contrarian claim, named person/company + action, open loop that forces the read-through

2. FACT GROUNDING (0-10)

Re-read data/research/<idea_id>.md. For EVERY factual claim in the draft (numbers, names, dates, quoted stats, product references), verify it traces to the brief. Uncited claims score 0 on this dimension. If the brief frontmatter has research_thin: true, allow claims from the idea's summary / transcript fields to pass. No inline citation markers are required from the writer (D-16) — you do the cross-check mentally.

Read the full file on GitHub · 110 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 · 110 lines · 48 tokens per session scan A 501d5f949415

Subscribe to this mod's changes

critic is an agent published in the GitHub repository robinsadeghpour/content-workflow (55 stars, last pushed 3mo ago), licensed MIT. It adds 48 tokens to every session and 1,556 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.