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.
git clone --depth 1 https://github.com/perniemann/pnCoreWrote 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/rules/perniemann/pncore/pn-tool-risk-policy)<a href="https://agentmods.dev/rules/perniemann/pncore/pn-tool-risk-policy"><img src="https://agentmods.dev/badge/rules/perniemann/pncore/pn-tool-risk-policy/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/rules/perniemann/pncore/pn-tool-risk-policy"><img src="https://agentmods.dev/badge/rules/perniemann/pncore/pn-tool-risk-policy.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 | $0.00480 | $0.00480 |
| Opus 5.5 | $0.00192 | $0.00192 |
| Sonnet 5.5 | $0.00096 | $0.00096 |
| Haiku 4.5 | $0.00048 | $0.00048 |
Grade A, and why
pn-tool-risk-policy 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 1mo 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.
What it actually says
Tool risk policy (pn-core MCP)
| Tier | Tools | Expectation |
|---|---|---|
| Read-only | list_*, get_*, health, project_context, harness_detect, list_workflow_types, suggest_model_tier, workflow_step, workflow_confirm, workflow_usage_totals, workflow_handoff_read, workflow_state_load, workflow_run_query |
No workspace writes; safe to call freely. |
| Append-only JSONL | report_usage, gate_log_append, workflow_handoff_append, workflow_state_save |
Writes under cwd only via path rules; prefer .pncore/ paths. Idempotent or append-only semantics. |
| Workspace scaffold | harness_scaffold |
Writes onboarding files only into the selected harness folders (.cursor/, .claude/, .agents/skills, .pi/, AGENTS.md managed block); workspace-contained; existing files skipped unless overwrite; dryRun returns the plan. Confirm harness selection with the user when detection is empty. |
| Verify (no shell) | workflow_verify |
Catalog argv only; requires disposeVerify; returns a GateReport. |
| Gate / secret | approval_checkpoint |
Requires PNCORE_APPROVAL_TOKEN on server; optional workflow_type + workflow_step + run_id for human-gate tickets. |
| External side effects | paperclip_issue_* |
Requires Paperclip env; treat as production API calls. |
Workflow human gates
- Optional mandatory human gates:
PNCORE_REQUIRE_APPROVAL_FOR_WORKFLOWS— each humanworkflow_stepneeds a freshapproval_checkpointticket. - Echo
run_idfromworkflow_stepin state for every call; pass the samerun_idtoapproval_checkpointwhen issuing tickets so tickets bind to the run.
Audit
- After gate outcomes, call
gate_log_appendwith optionalrun_idfor correlation with.pncore/workflow-runs.jsonl.
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.
- 1mo ago Changed · +2 lines · +124 tokens per session 74766bda84c9
- 1mo ago First seen · 23 lines · 356 tokens per session scan A 6fece9737274
pn-tool-risk-policy is a cursor rule published in the GitHub repository perniemann/pnCore (0 stars, last pushed yesterday), licensed MIT. It adds 480 tokens to every session, about $0.0019 per session on Opus 5.5. 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 2026-09-03.
Other cursor rules, from other repositories
t-800-mandatory-routing
A mandatory routing rule for creating or changing subagents, skills, commands, rules, and hooks. Hooks are scripts that run automatically when certain events occur.
t-800-operator-routing
A routing rule for explaining Cursor and its agent features to Russian-speaking beginners. Cursor is a code editor with built-in AI assistance.
t-800-factory-routing
A routing rule for building and coordinating groups of coding subagents. Subagents are separate agents that handle parts of a larger task.
t-800-cloud-hub-routing
A routing rule for Cloud Hub setup work, sending matching requests to a designated lead agent. Cloud Hub is the project area named in the rule.
typescript-conventions
Cursor rule "typescript-conventions" from VirtusLab-Open-Source/strapi-plugin-mcp, covering typescript and coding conventions, typescript configuration, import/export patterns, code organization and services (server/src/services/).
1001-typescript
APPLY TypeScript best practices WHEN writing code TO ensure type safety, readability, and maintainability.