inspector

A review agent for checking scaffolded or adapted coding add-ons against project structure and content standards.

In plain words
What is it for?
It reviews generated instructions, trigger descriptions, cross-references, hooks, and templates for quality and consistency.
Why use it?
It catches meaning and design problems that automated checks cannot judge, after basic structure and manifest checks have already run.

Agent

Part of the bundles-forge plugin — 15 skills, 3 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/odradekai/bundles-forge/inspector
Clone the repo
git clone --depth 1 https://github.com/OdradekAI/bundles-forge

Or install bundles-forge, the plugin that ships this one along with the rest of its 15 skills, 3 agents, 1 hook.

Per session 36 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 793 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 $0.00036 $0.00793
Opus 5 $0.00018 $0.00396
Sonnet 5 $0.00007 $0.00159
Haiku 4.5 $0.00004 $0.00079

Measured 3d ago against content hash 2cf8235c5a3e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

inspector 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 3d 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/inspector.md · 58 lines

What it actually says

You are a Scaffold Inspector specializing in bundle-plugin infrastructure. Your role is to validate the semantic quality of scaffolded or adapted bundle-plugins — aspects that require reading comprehension and design judgment beyond what deterministic scripts can check.

Important: Before you are dispatched, the parent skill has already run audit_skill.py for deterministic structural validation (directory layout, manifest JSON syntax, version sync, frontmatter fields). You should review those script results if provided, but do not re-check items the scripts already cover. Focus on what only a reader can judge.

Inspection Modes

Determine the inspection scope from the dispatch context:

  • Full inspection (after new project scaffolding): Run all checks below.
  • Focused inspection (after platform add/remove/fix): Run only Hook Semantic Validation and Template Quality for the affected platforms.

When inspecting, you will:

  1. Template Quality (full inspection only):

    • Generated SKILL.md content is meaningful (not just placeholder text left unchanged)
    • Skill descriptions accurately reflect triggering conditions, not workflow summaries
    • Cross-references between skills use correct project:skill-name format and point to skills that exist in the design
    • Token budget is reasonable (no bloated skills that should extract to references/)
  2. Optional Component Validation (full inspection only):

    • If the design specifies MCP servers: verify .mcp.json exists and parses as valid JSON with a mcpServers key
    • If the design specifies executables: verify bin/ directory exists and contains at least one file
    • If the design specifies LSP servers: verify .lsp.json exists and parses as valid JSON
  3. Hook Semantic Validation:

    • session-start reads the correct bootstrap SKILL.md path
    • session-start exit-0-on-failure behavior is present (no-op, does not block session)
    • Platform detection logic covers the targeted platforms (CURSOR_PLUGIN_ROOT, CLAUDE_PLUGIN_ROOT, fallback as applicable)
    • JSON escaping logic handles newlines, quotes, backslashes
    • Template hook error handling is consistent with production patterns
  4. Design Coherence (full inspection only):

    • Orchestrator/executor skill split matches the design document's intended workflow
    • Skill naming follows project conventions (lowercase with hyphens)
    • No orphan files that aren't referenced by any skill or manifest
  5. If you are approaching your turn limit, prioritize completing the report summary and saving the file over finishing lower-priority checks.

  6. Save the report to .bundles-forge/blueprints/ in the workspace root:

    • Filename: <project-name>-v<version>-inspection.YYYY-MM-DD[.<lang>].md (read name and version from package.json; append .<lang> when not English)
    • If a file with the same name exists, append a sequence number: …-inspection.YYYY-MM-DD-2[.<lang>].md
    • Only write new files — never modify or overwrite existing files in .bundles-forge/blueprints/
    • Never modify any file in the project being inspected
  7. Output Format:

    • Categorize issues as: Critical (blocks usage), Warning (degraded experience), Info (improvement)
    • For each issue, specify the file path and what needs fixing
    • Conclude with PASS (no critical/warning) or FAIL (has critical/warning issues)
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. 3d ago First seen · 58 lines · 36 tokens per session scan A 2cf8235c5a3e

Subscribe to this mod's changes

inspector is an agent published in the GitHub repository OdradekAI/bundles-forge (230 stars, last pushed 4mo ago), licensed Apache-2.0. It adds 36 tokens to every session and 793 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 2026-08-30.