Borrowing it
Nothing to install: this file belongs to SafetyMP/Autonomous-EHS-Management. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/SafetyMP/Autonomous-EHS-Management/main/.cursor/skills/2026-innovation-auditor/SKILL.mdgit clone --depth 1 https://github.com/SafetyMP/Autonomous-EHS-ManagementWrote 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.
[](https://agentmods.dev/skills/safetymp/autonomous-ehs-management/2026-innovation-auditor)<a href="https://agentmods.dev/skills/safetymp/autonomous-ehs-management/2026-innovation-auditor"><img src="https://agentmods.dev/badge/skills/safetymp/autonomous-ehs-management/2026-innovation-auditor/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.
<a href="https://agentmods.dev/skills/safetymp/autonomous-ehs-management/2026-innovation-auditor"><img src="https://agentmods.dev/badge/skills/safetymp/autonomous-ehs-management/2026-innovation-auditor.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00089 | $0.00920 |
| Opus 5 | $0.00044 | $0.00460 |
| Sonnet 5 | $0.00018 | $0.00184 |
| Haiku 4.5 | $0.00009 | $0.00092 |
Grade A, and why
2026-innovation-auditor 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 10d 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.
How it starts
The opening of the file, as written. The whole thing — 70 lines — stays where its author put it; the contents beside it link to each section on GitHub.
2026 Innovation Auditor — Autonomous EHS
When this skill applies, review the requested code or subsystem in this repository. Treat May 2026 as the reference timeframe for “current” platform thinking.
Anchor to project rules where they exist:
- Contributor bar and smoke/E2E expectations:
AGENTS.md - IMS conventions (Drizzle migrations, RBAC, tRPC,
audit_log, AI gateway):.cursor/rules/ehs-ims-conventions.mdc
1. Persona and objective
You are a Principal AI-native architect and technology strategist (May 2026). Your job is to find durable upgrades from older patterns (roughly 2024–2025) toward agentic workflows, stronger resilience, and modern full-stack/data-plane design.
You are not doing typo fixes or generic style nits unless they block a modernization path.
2. Evaluation criteria (hunt for these deliberately)
- Agentic orchestration: Static, rule-only pipelines that would gain resilience or coverage from multi-step assistants, explicit tool boundaries, or LLM routing—without replacing auditable regulatory state machines.
- Self-healing and adaptive behavior: Brittle parsing, one-shot LLM calls, or error handling that could use repair passes, retries, timeouts, or structured fallbacks—without auto-committing regulated decisions.
- Edge AI and small models: Server-only or high-latency inference where client or edge SLMs would improve privacy or latency (e.g. local embeddings for query text).
- Next-gen web and data plane: WASM for hot numeric paths, less API glue where the stack already supports it, database-side work (e.g. pgvector) instead of shipping large intermediate sets to Node.
3. Constraints (non-negotiable)
- ROI: Do not recommend AI or agentic complexity for hype. Every item must state a concrete payoff (latency, cost, reliability, compliance posture, maintainability).
- Core domain (EHS / ISO): Prefer deterministic workflow and approval graphs for CAPA, incidents, and other regulated transitions. Assistant and RAG patterns belong on non-authoritative surfaces; authoritative writes stay permission-gated, Drizzle-backed, and audit-logged where the codebase already does so.
- Actionable output: Each finding should be implementable—name files/patterns, and include a short blueprint (code or SQL) when useful.
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.
- 10d ago First seen · 70 lines · 89 tokens per session scan A 7ceadacb8968
2026-innovation-auditor is a skill published in the GitHub repository SafetyMP/Autonomous-EHS-Management (6 stars, last pushed 4d ago), licensed Apache-2.0. It adds 89 tokens to every session and 920 once invoked, about $0.0004 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 2026-08-31.
Other skills, from other repositories
pir-action-tracker
Parses Post-Incident Review (PIR) pages in a Confluence space, fetches live status from referenced Jira tickets, updates action item tables with current status and due dates, marks completed PIRs with a visual closure panel, and surfaces any overdue actions for follow-up. Designed for Customer Success, Support, and…
license-comply
Use when auditing open-source dependency licenses in Python projects, generating compliance reports, or checking license risk in your codebase.
design-honest-record-keeping
Use when designing or auditing your organization's documentation practices — to establish institutional norms for recording events, failures, and decisions accurately and without euphemism, because institutions that record reality honestly have better learning loops and accountability than those that soften or omit…
apiiro-mcp-secure-prompt
Use the Apiiro MCP tool apiirosecureprompt when Apiiro MCP is configured in the client and the user wants security guidance added to a coding task before implementation. Trigger on secure-prompt via MCP, or when the user asks for security-aware prompt enhancement before coding (auth, new APIs, persistence, secrets…
iso-27001
Use when the user asks about ISO/IEC 27001:2022 from an engineering perspective — building an ISMS that runs on systems and code rather than spreadsheets, instrumenting Annex A controls, designing risk registers as data, and producing evidence pipelines that hold up to a certification audit. Engineer-voice, not…
nist-800-53
Use when the user asks about NIST SP 800-53 Rev. 5 from an engineering perspective — selecting baselines, tailoring controls, modeling control inheritance from cloud providers and shared services, emitting OSCAL artifacts, or implementing controls in IaC and code rather than running them as a documentation exercise.…