manifest-curator

manifest-curator is an agent for Claude Code from V-Songbird/foundry. It costs 31 tokens per session (572 once invoked), scanned A, original, MIT.

Review the actual Claude or Codex manifests, catalogs and pinned revisions against their own schemas. Audit by default; report evidence and uncertainty.

Agent for Claude Code

Written for Claude Code: Claude Code plugin machinery. Also seen: model in frontmatter; mentions Codex.

Install with agentmods
npx agentmods add agents/v-songbird/foundry/manifest-curator
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.

Clone the repo
git clone --depth 1 https://github.com/V-Songbird/foundry

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 manifest-curator

README.md
[![agentmods](https://agentmods.dev/badge/agents/v-songbird/foundry/manifest-curator/github.svg)](https://agentmods.dev/agents/v-songbird/foundry/manifest-curator)
Your own site
<a href="https://agentmods.dev/agents/v-songbird/foundry/manifest-curator"><img src="https://agentmods.dev/badge/agents/v-songbird/foundry/manifest-curator/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for manifest-curator

Your own site · 80×15
<a href="https://agentmods.dev/agents/v-songbird/foundry/manifest-curator"><img src="https://agentmods.dev/badge/agents/v-songbird/foundry/manifest-curator.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 31 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 572 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 unknown 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 $0.00031 $0.00572
Opus 5 $0.00015 $0.00286
Sonnet 5 $0.00006 $0.00114
Haiku 4.5 $0.00003 $0.00057

Measured today against content hash 20f37c7ea3ec, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

manifest-curator 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 today.

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.

.claude/agents/manifest-curator.md · 57 lines

What it actually says

Manifest review for Foundry

Audit by default. Fix mode permits only the mechanical corrections explicitly within the task's scope; it does not authorize changing plugin behavior, publishing a release or rewriting unrelated local work.

Identify the plugin, edition, checkout and exact revision under review first. Foundry contains two different catalog formats:

  • Claude: .claude-plugin/marketplace.json; the entry owns the version and pins the Claude commit. The plugin metadata is .claude-plugin/plugin.json.
  • Codex: .agents/plugins/marketplace.json; the native .codex-plugin/plugin.json owns the version. The catalog carries its native interface/policy/category fields and pins the Codex commit.

Never translate one schema by changing the assistant's name. Do not demand a Claude owner block or catalog version from a Codex catalog. Do not remove the version required by a Codex plugin. Main selectors are not installable packages.

Read the authoritative references for the edition being audited:

If the live specification is unavailable, mark affected schema claims SPEC_UNVERIFIED. Do not invent a documentation URL or pass uncertain rules as verified. Use the current source and official specification rather than an old embedded description when they disagree.

Check JSON validity, identifiers, actual source paths, author consistency, supported fields, native component discovery, dependencies and description quality. In Foundry, inspect the implementation of scripts/git-hooks/check-platform-marketplaces.js and its result. Distinguish local ancestry/manifest validation from remote reachability and installed activation. A passing check of an old pin says nothing about uncommitted files.

Check versions against the selected edition's previous release. Both catalogs use a full validated source SHA with the correct platform ref. Do not add an unavailable Hush Codex edition merely to fill a symmetry gap.

Use .github/RELEASE_WORKFLOW.md for release ownership and sequencing. Preserve unrelated changes and any runtime freeze imposed by the user.

Return ERROR, WARNING, MANUAL and INFO findings with exact files/revisions and supporting evidence. Identify spec-dependent findings that remain unverified. Report what was actually changed only when fix mode was authorized; an audit returns findings without editing files.

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. today First seen · 57 lines · 31 tokens per session scan A 20f37c7ea3ec

Subscribe to this mod's changes

manifest-curator is an agent published in the GitHub repository V-Songbird/foundry (4 stars, last pushed today), licensed MIT. It adds 31 tokens to every session and 572 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 2026-09-10.

Related

Other agents, from other repositories

code-reviewer

Use this agent to review changed files for bugs, security issues, performance problems, and code quality. Use when reviewing PRs, checking code before committing, or auditing code quality. Examples: User: "Review the changes I made" Assistant: "I'll launch the code-reviewer agent to analyze your changes." User: "Is…

Sagargupta16/claude-skills · 96 tokens

guardrail-checker

Use this agent to verify that code changes follow development guardrails -- git safety, security practices, PR workflow, and coding standards. Examples: User: "Check if my changes follow our rules" Assistant: "I'll launch the guardrail-checker agent to validate compliance." User: "Audit this PR for best practice…

Sagargupta16/claude-skills · 89 tokens

pr-analyzer

Use this agent to analyze GitHub PRs -- check CI status, read review comments, assess merge readiness, and suggest next actions. Examples: User: "Check the status of my open PRs" Assistant: "I'll launch the pr-analyzer agent to check all your PR statuses." User: "Why is CI failing on my PR?" Assistant: "Let me launch…

Sagargupta16/claude-skills · 91 tokens

the-rfc-reviewer

Use to review engineering RFCs, design docs, technical proposals, or architecture write-ups the way a senior staff engineer would. Trigger when the user shares a doc and asks for review, an EM wants a second opinion before approving a proposal, or the user wants structured critique before publishing their own RFC.

shrwnsan/vibekit-claude-plugins · 66 tokens

the-explainer

Use to write explanations of how an AI feature works — sales enablement, customer docs, exec briefings, support runbooks, FAQ, security/legal review materials. Trigger when the user says "I need to explain X to [audience]", "write a how-it-works doc", "draft an FAQ for…", or when prepping launch enablement. Calibrates…

shrwnsan/vibekit-claude-plugins · 93 tokens

the-prompt-critic

Use to review production prompts, system prompts, or agent instructions the way a senior engineer reviews code. Trigger when the user shares a prompt and asks "is this good?", when iterating on a struggling LLM feature, or proactively before any prompt ships to production.

shrwnsan/vibekit-claude-plugins · 59 tokens