claude-md-progressive-disclosurer

An agent skill by daymade, from daymade/claude-code-skills. Tags: code-quality, documentation, planning, review, strategy.

What it does

Optimize, slim, or restructure CLAUDE.md/AGENTS.md with progressive disclosure and zero information loss. Use when the user explicitly asks to audit, 精简, 瘦身, 重构, split, or diagnose adherence problems in instruction files. Profiles the whole resident startup surface, allocates rules among prose, path rules, Skills, hooks, and references, then moves low-frequency sections verbatim with content-integrity checks. Also use when an active task starts moving or compressing instruction sections. Not for generic task drift unless instruction files are in scope.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add daymade/claude-code-skills --skill claude-md-progressive-disclosurer

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/daymade/claude-code-skills
cp -r claude-code-skills/daymade-claude-code/claude-md-progressive-disclosurer ~/.claude/skills/claude-md-progressive-disclosurer

Safety box score

Grade C, moderate risk: 28 of 100 risk points, rated 2026-10-18. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill is a local Markdown rewriter/auditor for `CLAUDE.md`/`AGENTS.md` plus helper docs and Python scripts. Its main risks are broad file-write reach, reading home-directory config files that may contain credentials, and optional web lookups for external reference material; it does not include privilege escalation or opaque binaries.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	2 (moderate)	Two shipped Python CLIs (`scripts/profile_claude_md.py`, `scripts/sink_sections.py`) are directly runnable and perform file I/O, but they do not eval external code or spawn arbitrary shell pipelines.
Network access	1 (low)	The docs explicitly tell the agent to refresh by checking external sources and contain multiple public documentation URLs, but the code itself has no fetch/send logic.
Filesystem reach	2 (moderate)	The sink tool rewrites a `source` path and appends to per-section `target_ref` paths from a JSON spec, and the docs also direct reads of home-dotfile config locations like `~/.claude/...` and `~/.codex/...`.
Secrets & credentials	2 (moderate)	The instructions explicitly inspect `~/.claude/settings.json` and `~/.codex/config.toml`, which can expose tokens or private account/config data to the model.
Privilege & persistence	0 (none)	No sudo, global installation, service registration, or shell/OS setting changes appear in the scripts or docs.
Instruction hijack surface	0 (none)	I do not see hidden text or instructions to ignore safeguards, follow untrusted remote prompt text, or otherwise subvert higher-priority instructions.
Irreversible actions	0 (none)	The skill does not push, deploy, message, pay, book, or delete infrastructure; its actions stay within local file auditing and rewriting.
Transparency	0 (none)	All shipped artifacts are plain-text Markdown and Python; there are no binaries, base64 blobs, or unpinned downloaded installers.

Before installing

Read `scripts/sink_sections.py` and confirm you are comfortable with a spec that can rewrite arbitrary regular-file paths.
Check whether you are okay exposing `~/.claude/settings.json` and `~/.codex/config.toml` to the agent during verification.
Run the scripts once in a scratch repo to observe the backup and overwrite behavior before using them on real instruction files.

Rated 2026-10-18 by an automated review of 7 files from the skill's repository.

Source

Repository: daymade/claude-code-skills (all skills from this repository)
Path: daymade-claude-code/claude-md-progressive-disclosurer/SKILL.md
Branch: main
Updated: 2026-09-20

grill-with-docs — A relentless interview to sharpen a plan or design, which also creates docs (ADR's and glossary) as we go.
marketing-plan — When the user needs a comprehensive marketing plan for a client, a company they advise, or their own product.
principle-minimize-reader-load — Apply when reviewing or shaping code that's hard to trace.
claude-migrate-memory-to-doc — Migrates Claude Code personal memory (the per-project memory/ directory) into tool-agnostic reference docs.
docs-cleaner — Consolidates redundant documentation while preserving all valuable content; keeps a doc set truthful after things change.
skill-reviewer — Reviews and improves Claude Code skills against official best practices.