dramaturg

dramaturg is an agent for Claude Code from kbatsu/chrysippus. It costs 47 tokens per session (660 once invoked), scanned A, original, MIT.

A checking agent that reviews whether a conversation is following the rules of an active persona, such as a particular speaking style or set of safety limits. A persona is a role with instructions that shape how an agent responds.

In plain words
What is it for?
Use it to audit an agent response or conversation, investigate persona behavior, and check whether style and guardrail requirements were followed.
Why use it?
It helps find why a persona is drifting, breaking its rules, or responding in the wrong style. It can also examine a conversation and identify specific violations.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the chrysippus plugin — 5 skills, 6 commands, 6 agents, 1 hook shipped together

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/kbatsu/chrysippus/dramaturg
Clone the repo
git clone --depth 1 https://github.com/kbatsu/chrysippus

Made for: Claude Code.

Or install chrysippus, the plugin that ships this one along with the rest of its 5 skills, 6 commands, 6 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 dramaturg

README.md
[![agentmods](https://agentmods.dev/badge/agents/kbatsu/chrysippus/dramaturg.svg)](https://agentmods.dev/agents/kbatsu/chrysippus/dramaturg)
Your own site
<a href="https://agentmods.dev/agents/kbatsu/chrysippus/dramaturg"><img src="https://agentmods.dev/badge/agents/kbatsu/chrysippus/dramaturg.svg" alt="Measured on agentmods" height="20"></a>
Per session 47 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 660 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 5.1 $0.00047 $0.00660
Opus 5 $0.00023 $0.00330
Sonnet 5 $0.00009 $0.00132
Haiku 4.5 $0.00005 $0.00066

Measured 6d ago against content hash 7e20aa991235, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

dramaturg 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 6d 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.

agents/dramaturg.md · 76 lines

How it starts

The opening of the file, as written. The whole thing — 76 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the dramaturg — a meta-agent that audits persona-skill adherence. You do not adopt any persona voice yourself; you speak plainly and analytically.

When to invoke

  • A user suspects the active persona is drifting from its rules (wrong register, violating guardrails, ignoring preservation).
  • Debugging why a persona's output doesn't match its SKILL.md.
  • Auditing a conversation transcript for rule-adherence.
  • Checking that the stereotype-drift guardrails (esp. pirate §7, toronto-mans §7, gen-alpha §7) are being honoured.

Input

The user provides one of:

  • A recent assistant response to audit (pasted in).
  • A conversation transcript.
  • A specific claim ("the active persona should have yielded to plain English here but didn't").

The user should also say which persona is (or should be) active. If not specified, infer from the transcript; if ambiguous, ask.

Audit checklist

For the specified persona, read the relevant files in order:

  1. rules/<persona>/_meta.json — flavors, preserve toggles, triggers.
  2. rules/<persona>/instructions.md — full register rules.
  3. rules/<persona>/lexicon.md (if present) — vocabulary scope.
  4. rules/<persona>/examples.md — worked examples and anti-examples.

Then check the subject content for:

  • Register correctness: is the prose in the right voice for the active flavor? Cite specific phrases that match/miss.
  • Preservation compliance: do backtick contents, commits, safety warnings, error text, code comments stay plain English per the <persona>.config defaults? Cite violations.
  • Guardrails: are any explicit prohibitions (excluded lexicon, forbidden patterns, stereotype-drift rules) being violated? Cite.
  • Flavor consistency: does the flavor stay consistent across the response, or does it drift?
  • Appropriate yield-to-plain: for safety warnings / errors / items that require verbatim preservation, did the skill correctly yield the floor?

Output format

Read the full file on GitHub · 76 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. 6d ago First seen · 76 lines · 47 tokens per session scan A 7e20aa991235

Subscribe to this mod's changes

dramaturg is an agent published in the GitHub repository kbatsu/chrysippus (3 stars, last pushed 1mo ago), licensed MIT. It adds 47 tokens to every session and 660 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

lean-crew-investigator

Read-only code locator. Returns file:line table for "where is X defined", "what calls Y", "list all uses of Z", "map this directory". Output is lean-compressed so the main thread eats 60% fewer tokens than vanilla Explore. Refuses to suggest fixes.

utk2103/Prompt-Studio · 65 tokens

quality-fixer

Specialized agent for verifying software projects and fixing quality failures within the current task scope. Use proactively after code changes or for quality, test, build, lint, format, correctness, or fix requests.

shinpr/claude-code-workflows · 44 tokens

developer

Use when execute-round's Phase 3 (dev body) needs to implement BA design exactly. Writes source + tests per file decomposition, runs pre-audit quality gates, registers forward-debts, and reports diff summary.

Arch1eSUN/Arcgentic · 47 tokens

doctrine-performance-optimizer

Read-only performance audit of Doctrine usage: N+1 queries, fetch modes, batch processing, missing indexes, and caching opportunities. Use proactively after adding entities, relations, repository queries, or when a page/endpoint is reported slow.

dev-toolings/superpowers-symfony · 53 tokens

arcgentic-auditor

Dispatched when a round is in auditinprogress state. Produces a verdict file at the project's auditsdir following the canonical 9-section template, with a mechanically-verifiable fact table, structured findings, and lesson-codification result. Does NOT read planner/developer reasoning chains — audit independence is…

Arch1eSUN/Arcgentic · 101 tokens

sniper

Use when: after ANY code modification (mandatory post-edit validation). Do NOT use for: new features, quick fixes already identified (use sniper-faster), read-only analysis.

fusengine/agents · 38 tokens