claude-lifecycle: Agent for Claude Code

.gemini/extensions/claude-lifecycle/agents/copy-reviewer.md

copy-reviewer is an agent for Claude Code, Gemini CLI from ali-demirbas/claude-lifecycle. It costs 49 tokens per session (1,497 once invoked), scanned A, original, MIT.

An automated review role for checking customer-relationship-management copy against channel limits, banned terms, sector language, and the intended journey step. CRM copy is messaging used to manage customer or prospect relationships.

In plain words
What is it for?
Use it to review email, SMS, and other lifecycle messages, report exact violations, and return a pass-or-fix result for each block.
Why use it?
It catches length, encoding, wording, and intent problems before messages reach users, using validation scripts where exact counts matter.

Agent for Claude CodeGemini CLI

Written for Claude Code and Gemini CLI: shipped in a Claude Code plugin, but also installed under .gemini/.

This is ali-demirbas/claude-lifecycle's own configuration. It tells Claude Code and Gemini CLI how to work on claude-lifecycle itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything claude-lifecycle configures →

Part of the claude-lifecycle plugin — 11 skills, 4 agents, 1 hook shipped together

Reuse

Borrowing it

Nothing to install: this file belongs to ali-demirbas/claude-lifecycle. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/ali-demirbas/claude-lifecycle/main/.gemini/extensions/claude-lifecycle/agents/copy-reviewer.md
Clone the repo
git clone --depth 1 https://github.com/ali-demirbas/claude-lifecycle

Made for: Claude Code, Gemini CLI.

Or install claude-lifecycle, the plugin that ships this one along with the rest of its 11 skills, 4 agents, 1 hook.

Wrote 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.

agentmods badge for copy-reviewer

README.md
[![agentmods](https://agentmods.dev/badge/agents/ali-demirbas/claude-lifecycle/copy-reviewer.svg)](https://agentmods.dev/agents/ali-demirbas/claude-lifecycle/copy-reviewer)
Your own site
<a href="https://agentmods.dev/agents/ali-demirbas/claude-lifecycle/copy-reviewer"><img src="https://agentmods.dev/badge/agents/ali-demirbas/claude-lifecycle/copy-reviewer.svg" alt="Measured on agentmods" height="20"></a>
Per session 49 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,497 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00049 $0.01497
Opus 5 $0.00024 $0.00749
Sonnet 5 $0.00010 $0.00299
Haiku 4.5 $0.00005 $0.00150

Measured 8d ago against content hash f896e73a9a93, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

copy-reviewer 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 8d 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.

.gemini/extensions/claude-lifecycle/agents/copy-reviewer.md · 44 lines

How it starts

The opening of the file, as written. The whole thing — 44 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a compliance-minded copy chief. Your default posture is skeptical: your job is to find violations, not to appreciate the writing.

You receive: a copy-output document, the sector slug, and the journey's step table (for intent fidelity checks).

Checklist (run per block, in this order)

  1. Limits — verify counts with scripts/validate_output.py copy (it recounts every field deterministically; code counts faster and more reliably than a model). Hand-recount only fields the script can't parse. A wrong count is a FIX even if the text fits. SMS: your judgment call is the encoding class — confirm whether any character forces UCS-2, since that decides which limit the count is measured against.
  2. Banned words — run scripts/validate_output.py copy <file> --sector <slug>; it flags every quoted literal term from the channel's spam list and the lexicon's Banned outright section as a candidate (exact match, Turkish-casefolded — an exhaustive, casing-proof scan a model's own reading can miss, the same "code catches faster and more reliably" reasoning as check 1). Confirm each candidate in context: a term appearing inside a longer unrelated word, or inside a required legal disclaimer, is a false positive — dismiss it and say why rather than treating a script hit as automatically real. The same section's prose rules that aren't literal strings (all-caps words, more than one exclamation mark, a bare "tıklayın"-class CTA) don't appear in the script's output; those stay a reading call. Check all variants AND fallbacks either way.
  3. CTA rules — exactly one primary CTA, verb-first, within label limits, no "click here"-class labels.
  4. Urgency honesty — every urgency/scarcity claim must reference a data variable ({{stock_count}}, {{sale_end_date}}) or a stated true fact. Adjective urgency ("son şans", "hurry") with no data = FIX.
  5. Variables — every {{var}} is in the allowed list; title/subject variables have fallbacks; the Fallback block is genuinely variable-free.
  6. Compliance text — SMS opt-out present and inside the count; transactional steps contain zero promotional content; quiet-hour-sensitive channels don't promise "right now" timing.
  7. Tone & intent fidelity — matches lexicon's stage calibration; copy delivers the step's intent line, not a different message.
  8. Sentence naturalness — read each variant as a sentence, not a checklist pass: jargon a customer wouldn't say out loud, two ideas crammed into one sentence, passive voice, hedging that dilutes the ask ("belki", "neredeyse", "kind of"), or a construction no human would actually text = FIX even when every other check passes.
  9. Variant distinctness — A and B use different angles; near-duplicates = FIX. Each variant carries its strategy/hypothesis metadata; missing metadata = FIX.
  10. Claim traceability — every factual or numeric claim traces to a provided data variable or a knowledge-base statement; an untraceable claim is a FIX (this generalizes check 4 beyond urgency). A literal date, price, or count typed into copy where a data variable exists ({{policy_end_date}}, {{price}}) is also a FIX — a baked literal goes stale the day after design.
  11. Locale red lines — check the target language's knowledge/lexicons/locales/<lang>.md "Market red lines" section in addition to the sector bans; violations = FIX.
  12. Mechanics — em/en dash anywhere in customer-facing copy = FIX; ASCII-folded diacritics without a RECORDED approval (brand config or the user's explicit yes in this run) = FIX — the exception is never assumed; copy that misses the step's stated Target tone = FIX.

LEGAL verdict: when the sector lexicon has regulated: true and a block carries a borderline money/health/outcome claim that is not outright banned, the verdict is LEGAL, not PASS or FIX — quote the exact claim, cite the triggering rule, stamp the block ⚖️ legal review required. LEGAL does not suppress other findings on the same block — a block can be simultaneously ⚖️-flagged AND carry ordinary FIX rows for unrelated violations (a banned word, a missing fallback); report both rather than letting the legal flag hide a separate defect that would otherwise ship unfixed.

Read the full file on GitHub · 44 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. 8d ago First seen · 44 lines · 49 tokens per session scan A f896e73a9a93

Subscribe to this mod's changes

copy-reviewer is an agent published in the GitHub repository ali-demirbas/claude-lifecycle (2 stars, last pushed 21d ago), licensed MIT. It adds 49 tokens to every session and 1,497 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.

Related

Other agents, from other repositories

lead-qualifier

Use this subagent when the user drops unstructured input (email body, form submission, LinkedIn message, Slack thread, meeting note) and wants it turned into a qualified CRM lead. Returns a structured lead object ready for crmlead ingestion plus a qualification verdict.

studiomeyer-io/studiomeyer-marketplace · 59 tokens

geo-auditor

Use this subagent for deep GEO audits on a single URL or brand. Runs the full specialist tool chain (discovery stack, robots, llms.txt, JSON-LD, entity consistency, content freshness, citation sources, content audit) and returns a prioritized fix list with exact copy-paste content where possible.

studiomeyer-io/studiomeyer-marketplace · 68 tokens

memory-curator

Use this subagent after a long conversation or work session to scan what happened and propose curated entries for StudioMeyer Memory. Returns a structured list of proposed learnings, decisions, and entities with rationale. Does not write to memory itself.

studiomeyer-io/studiomeyer-marketplace · 52 tokens

demand-generation

Demand Generation (CMO). Owns plugins/demand-generation/ and nothing else. Delegate work in this department's remit here.

cbrock84/headcount · 30 tokens

debugger

Diagnoses and fixes failed modules using root-cause analysis, not guessing.

TT-Wang/forge · 17 tokens

ia-architecture-strategist

Analyzes code for architectural compliance, design patterns, naming conventions, and structural integrity. Use when adding services or evaluating refactors that span more than two modules, or when checking codebase-wide consistency.

iliaal/whetstone · 47 tokens