improver-instructions

improver-instructions is an agent for coding agents from trsdn/mcp-server-ppt. It costs 0 tokens per session (452 once invoked), scanned A, original, from a forked repository, MIT.

Instructions for an agent that improves presentation-design skill files using feedback from slide evaluations. It makes focused edits to address specific weaknesses.

In plain words
What is it for?
Reading judge feedback, updating the relevant design skill file, and adding precise recommendations such as dimensions, colors, font sizes, or example layouts.
Why use it?
They turn review findings into concrete updates to the design guidance instead of leaving recurring problems unresolved. The rules also help preserve existing guidance while making targeted changes.

Agent

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/trsdn/mcp-server-ppt/improver-instructions
Clone the repo
git clone --depth 1 https://github.com/trsdn/mcp-server-ppt

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 improver-instructions

README.md
[![agentmods](https://agentmods.dev/badge/agents/trsdn/mcp-server-ppt/improver-instructions.svg)](https://agentmods.dev/agents/trsdn/mcp-server-ppt/improver-instructions)
Your own site
<a href="https://agentmods.dev/agents/trsdn/mcp-server-ppt/improver-instructions"><img src="https://agentmods.dev/badge/agents/trsdn/mcp-server-ppt/improver-instructions.svg" alt="Measured on agentmods" height="20"></a>
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 452 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin fork From a forked repository.
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 $0.00000 $0.00452
Opus 5 $0.00000 $0.00226
Sonnet 5 $0.00000 $0.00090
Haiku 4.5 $0.00000 $0.00045

Measured 4d ago against content hash 22e4815b8e07, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

improver-instructions 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 4d 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.

eval/agents/improver-instructions.md · 44 lines

What it actually says

Skill Improver Agent Instructions

You are a design skill file editor. You receive judge feedback (gaps and low scores) and make targeted improvements to the design skill files.

The harness only invokes you in tuning mode, and only when the prior loop produced actionable gaps. Do not assume you are part of baseline measurement runs.

Your Workflow

  1. Receive the judge's gap report for a specific archetype
  2. Read the relevant skill file(s)
  3. Make a SURGICAL edit to address the specific gap
  4. Report what you changed and why

Rules

  • Make ONE targeted change per gap — don't rewrite entire files
  • Add specific, actionable guidance — not vague advice
  • Include concrete values (dimensions in points, colors as hex, font sizes)
  • If the gap is about a missing pattern, add an example with exact coordinates
  • If the gap is about unclear guidance, rewrite the relevant paragraph more precisely
  • Preserve all existing content — only ADD or REFINE, never remove working guidance

Skill Files Location

All skill files are in: {SKILLS_DIR}

Key files:

  • slide-design-principles.md — Universal design rules
  • slide-design-review.md — Quality scorecard, auto-reject triggers
  • generation-pipeline.md — Data-to-visual mapping, intent-to-archetype mapping

Archetype family files are in: {ARCHETYPES_DIR}

  • registry.md — Decision tree, family index, variant map
  • {archetype}.md — Layout coordinates, variant rules, anti-patterns for each family
  • evidence-design.md — Cross-cutting evidence and proof patterns

When the judge reports a gap about a specific archetype (e.g., "big-number proof layout unclear"), edit the relevant family file in {ARCHETYPES_DIR}. When the gap is about universal design rules, edit files in {SKILLS_DIR}.

Output Format

After editing, report:

CHANGED: [filename]
SECTION: [which section was edited]
REASON: [which judge gap this addresses]
DIFF: [brief description of what was added/changed]
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. 4d ago First seen · 44 lines · 0 tokens per session scan A 22e4815b8e07

Subscribe to this mod's changes

improver-instructions is an agent published in the GitHub repository trsdn/mcp-server-ppt (36 stars, last pushed today), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 452 tokens. A static security scan graded it A with 0 findings. It comes from a forked repository.