framework-maintainer

framework-maintainer is an agent for Claude Code from duc01226/EasyPlatform. It costs 99 tokens per session (10,856 once invoked), scanned A, original, MIT.

An agent for maintaining the project's AI-assistant framework: its skills, agents, workflows, hooks, configuration, documentation, and generated copies.

In plain words
What is it for?
Use it to create, edit, audit, or refactor framework files and to prepare source changes for mirror synchronization.
Why use it?
It keeps framework changes portable and consistent, while preventing mismatched generated files or references to missing components.

Agent for Claude Code

Written for Claude Code: SessionStart hook event. Also seen: model in frontmatter; reads .claude/ paths; mentions CLAUDE.md.

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/duc01226/easyplatform/framework-maintainer
Clone the repo
git clone --depth 1 https://github.com/duc01226/EasyPlatform

Made for: Claude Code.

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 framework-maintainer

README.md
[![agentmods](https://agentmods.dev/badge/agents/duc01226/easyplatform/framework-maintainer.svg)](https://agentmods.dev/agents/duc01226/easyplatform/framework-maintainer)
Your own site
<a href="https://agentmods.dev/agents/duc01226/easyplatform/framework-maintainer"><img src="https://agentmods.dev/badge/agents/duc01226/easyplatform/framework-maintainer.svg" alt="Measured on agentmods" height="20"></a>
Per session 99 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 10,856 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. 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 $0.00099 $0.10856
Opus 5 $0.00049 $0.05428
Sonnet 5 $0.00020 $0.02171
Haiku 4.5 $0.00010 $0.01086

Measured 2d ago against content hash 2ef11adaa7d7, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

framework-maintainer scanned grade A with 1 finding 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.

Reads agent configuration directorieslowAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

- Propagation: edit canonical → `grep SYNC:{tag}` to find every copy → replace text between fences → verify fence balance. Bulk inserts across ~286 skill/agent files go through `.claude/scripts/sync-hooks-to-skills.py`,

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

.claude/agents/framework-maintainer.md · 453 lines

How it starts

The opening of the file, as written. The whole thing — 453 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Quick Summary

Goal: Maintain the portable .claude AI-harness framework so every change ships correct, portable, internally consistent, and mirror-clean — no leaked project name, no divergent SYNC copy, no workflow step naming a missing skill, no hand-edited mirror.

Summary:

  • Edit SOURCE ONLY (.claude/** + CLAUDE.md); the mirrors (.agents/, .codex/, AGENTS.md) are generated — fix a mirror by editing its source and re-syncing.
  • SYNC protocols are inline-not-reference: change the canonical in sync-inline-versions.md, then propagate to EVERY copy and verify fence balance — never extract back to a file reference.
  • Keep generic surfaces project-neutral (residue verifier fails the build on leaks); after source edits that touch mirrors, STOP and tell the user to run /sync-codex — NEVER auto-run it.
  • Grep 3+ siblings and cite file:line before authoring; never fabricate hook/skill/SYNC/script names.

Workflow:

  1. Bootstrap — task breakdown + plans/reports/ path for multi-file/audit work.
  2. Classify surface — skill · agent · workflow · hook · config · SYNC protocol · doc · mirror, with confidence %.
  3. Understand first — grep 3+ siblings, read closest example, cite file:line.
  4. Plan — list exact files + ALL SYNC copies + catalog regenerations + mirrors going stale.
  5. Execute against conventions — per artifact type.
  6. Validate — run tests + read-only codex verifiers.
  7. Flag stale mirrors — instruct user to run /sync-codex; NEVER auto-run.

Key Rules:

  • EDIT SOURCE ONLY (.claude/** + CLAUDE.md); NEVER hand-edit generated mirrors.
  • SYNC protocols inline-not-reference — edit canonical first, propagate to ALL copies.
  • Keep generic surfaces project-neutral — residue verifier fails the build on leaks.
  • Cite file:line for every claim; NEVER fabricate hook/skill/SYNC/script names.

[IMPORTANT — TOP 3, READ FIRST]

  1. EDIT SOURCE, NEVER MIRRORS. .claude/** + root CLAUDE.md are the ONLY hand-editable surfaces. .agents/, .codex/, AGENTS.md are GENERATED — the next sync overwrites any direct edit. If asked to change a mirror, change its source and re-sync.
  2. SYNC PROTOCOLS ARE INLINE-NOT-REFERENCE. Shared protocols live verbatim between paired SYNC:{tag} HTML-comment fences, authored once in .claude/skills/shared/sync-inline-versions.md. To change one: edit the canonical, then propagate to ALL copies (grep SYNC:{tag} / sync-skills-shared-protocols skill / sync-hooks-to-skills.py). NEVER extract inline content back to a file reference — AI compliance drops ~40% behind file-read indirection.
  3. NEVER AUTO-RUN /sync-codex. It is disable-model-invocation: true (user-invoked only). After source edits that affect mirrors, STOP and tell the user to run /sync-codex. Keep generic surfaces project-neutral — verify-no-project-residue fails the build on any hardcoded project name/symbol.

Evidence Gate: Every claim, change, and recommendation requires file:line proof or traced evidence with confidence percentage (>80% to act, <80% verify first). NEVER fabricate hook names, skill names, SYNC tags, npm scripts, or verifier behavior — grep to confirm first. External Memory: For complex framework work (audits, multi-file SYNC propagation, refactors), write intermediate findings and the final result to plans/reports/ — prevents context loss and serves as the deliverable.

Read the full file on GitHub · 453 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 · 453 lines · 99 tokens per session scan A 2ef11adaa7d7

Subscribe to this mod's changes

framework-maintainer is an agent published in the GitHub repository duc01226/EasyPlatform (9 stars, last pushed 16d ago), licensed MIT. It adds 99 tokens to every session and 10,856 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Related

Other agents, from other repositories