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/jcooley8/pincushion-pluginWrote 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/jcooley8/pincushion-plugin/critique-latest-deploy)<a href="https://agentmods.dev/rules/jcooley8/pincushion-plugin/critique-latest-deploy"><img src="https://agentmods.dev/badge/rules/jcooley8/pincushion-plugin/critique-latest-deploy/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/jcooley8/pincushion-plugin/critique-latest-deploy"><img src="https://agentmods.dev/badge/rules/jcooley8/pincushion-plugin/critique-latest-deploy.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 | $0.00011 | $0.01288 |
| Opus 5.5 | $0.00004 | $0.00515 |
| Sonnet 5.5 | $0.00002 | $0.00258 |
| Haiku 4.5 | $0.00001 | $0.00129 |
Grade A, and why
critique-latest-deploy 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 yesterday.
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 — 102 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Call get_pending_critiques first. Keep a row with no deploymentId on the
legacy path; never make up deployment provenance.
For a deployment row, union its registered root/inventory with bounded rendered same-origin, robots, and nested-sitemap discovery. If provider route hints are absent, probe this union rather than labeling all pages changed. Reuse only local origin-bound auth and run:
node <pincushion-mcp>/scripts/capture-snapshot.mjs --reconcile "<root-url>" "<temporary-output-dir>" --inventory=<projectPageInventory-json-or-file> --previous-receipts=<latestCaptureReceipts-json-or-file>
The helper performs desktop 1280×900 plus mobile 390×844 settled captures,
masks transient tiles, and writes a current-run pair for unchanged as well as
added/persistent changes during a fresh frame run.
Missing links do not remove pages; terminal 404/410 or verified out-of-scope
redirects do. Login redirects, timeout, denied access, and invalid current
evidence are inaccessible/failed, not unchanged. A committed legacy receipt
without a valid v1 fingerprint is bootstrap_required when current
desktop+mobile evidence is valid: stage it as updated. Stop on truncated discovery.
Before staging or reporting, call register_deployment_discovered_pages with
the exact bounded reconciliation.discoveredPageUrls list.
Use only returned acceptedPendingPages / acceptedPageUrls; never upload or
report a locally discovered URL that was not accepted pending.
For each accepted pending non-removed terminal result, call record_deployment_page_result. First stage both newly staged desktop/mobile exact capture files with upload_page_snapshot,
including deploymentId, returned-byte contentHash, and captureFingerprint;
require immutable version IDs and staged: true. For unchanged, the server
accepts only byte hashes exactly matching the frozen canonical pair, records
observed evidence, and discards private duplicates; use updated when either
hash differs. Only deploy-hook-frozen removed rows are excluded; do not submit
removed for an initialized local run. Inaccessible/failed results remain
truthful non-current outcomes. The backend atomically promotes staged captures
and the canonical capture report after all terminal results.
Retry an uncertain acknowledgement only with the exact same
record_deployment_page_result body; returned idempotent: true is success.
After a completed reconciliation receipt, Crit only changed/added pages and mint
an independent immutable Crit receipt. Do not call complete_critique_request
before reconciliation and the eligible Crit receipt are committed. The
snapshot-latest-deploy compatibility alias uses this same checkpoint and never
drains the queue independently.
For an authenticated target, supply PINCUSHION_STORAGE_STATE created for the
exact origin and pass --require-auth. Supplied or required auth that is
missing, unreadable, symlinked, non-regular, insecure, malformed, unbound,
rejected with 401/403, redirected to auth, or shown as a signed-out shell is a
hard non-success: do not stage, upload, or record it as current. Require receipt
auth.mode: "authenticated"; with no supplied/required state, public capture
remains auth.mode: "anonymous". State contents stay local.
Create or refresh that local pair with npx pincushion-mcp snapshot --login "<root-url>" --proof-selector '<app-specific CSS selector visible only when signed in>'. The owner-only proof sidecar is bound to the exact origin and
state bytes; its selector must be visible on every authenticated route. Missing
or non-visible proof (including a neutral signed-out 200 shell) exits 6. The
initial URL and every redirect are pre-checked; a private/reserved DNS/IP answer
is rejected before navigation.
Fresh frame run
Frame freshness is independent of autoCritique: call
get_pending_deployment_frame_refreshes even when get_pending_critiques is
empty. The latter controls only whether Crit is requested. Use any first
reconciliation only to discover scope—do not upload or terminalize it. Register
that bounded discovery with register_deployment_discovered_pages; then call
start_deployment_capture_run. Keep its captureProof only in the
authenticated MCP caller's memory—never in shell argv, a command, report, pin,
status, or file. Re-run the fresh browser pass with only the safe run ID:
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.
- yesterday First seen · 102 lines · 11 tokens per session scan A 13ae7b9b068f
critique-latest-deploy is a cursor rule published in the GitHub repository jcooley8/pincushion-plugin (1 stars, last pushed yesterday), licensed MIT. It adds 11 tokens to every session and 1,288 once invoked, about $0.0000 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 2026-11-03.
Other cursor rules, from other repositories
prototype-previewer
Build interactive prototype reviewers with synced review notes and Figma capture pages.
ant-design
A set of guidelines for using Ant Design, a collection of ready-made interface components, in React applications. It covers code organization, performance, security, testing, and common mistakes.
7-day-nextjs-web-design-cursor
A seven-day design workflow for web products that will be built in Next.js.
ponytail
Ponytail, lazy senior dev mode. Always pick the simplest solution that works.
angular-20
This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.
dev-standard
Apache Superset development standards and guidelines for Cursor IDE.