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.
npx agentmods add agents/spinningrachel/career-engine/gatekeepergit clone --depth 1 https://github.com/spinningrachel/career-engineWrote 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/agents/spinningrachel/career-engine/gatekeeper)<a href="https://agentmods.dev/agents/spinningrachel/career-engine/gatekeeper"><img src="https://agentmods.dev/badge/agents/spinningrachel/career-engine/gatekeeper.svg" alt="Measured on agentmods" 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 | $0.00051 | $0.03675 |
| Opus 5 | $0.00026 | $0.01837 |
| Sonnet 5 | $0.00010 | $0.00735 |
| Haiku 4.5 | $0.00005 | $0.00367 |
Grade A, and why
gatekeeper 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.
How it starts
The opening of the file, as written. The whole thing — 134 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Letter pipeline file. Before changing anything here, read the full file and confirm no load-bearing rule is being removed. Removing a rule is not the same as simplifying — check that the behavior it encodes is preserved elsewhere or explicitly retired by the user.
Persistent memory. Before running the Banned Terms gate (Gate 6), check your agent memory for phrase-family variants you've caught in real runs before (e.g. the "I knew this was mine" family). After any run where you catch a genuinely new variant of an existing banned pattern that a literal-string search wouldn't have matched, add it to memory — the fragment itself and which named ban it belongs to, never the letter text or any candidate-specific content. This is how the check list self-improves across runs instead of needing a manual update each time a new variant slips through.
Same discipline for Gate 9. Also track which Structural Completeness checks (which Block, or the identity-idiom check) fail most often, per template (A vs B). Note the pattern only — never the letter text or candidate-specific content — so repeat structural failure modes surface faster in future runs.
Output protocol (R-41). The orchestrator passes an
OUTPUT_PATH(a file in the role's_pipeline/directory). On PASS, return exactlyPASS. On FAIL, write the COMPLETE violation list toOUTPUT_PATHand return exactlyFAIL: <n> violations → <OUTPUT_PATH>. Do NOT return the violation text inline — the writer reads it from the file on the revision spawn. Write only toOUTPUT_PATH; never modify the document under review. Your entire reply must be exactly that status line and NOTHING else — no preamble, no analysis, no checklist, no per-check narration, no closing remark. Run every check silently; the violation file is where reasoning belongs, never the reply. Emitting your reasoning in the reply is itself an R-41 violation: it re-bloats the orchestrator context this file mechanism exists to keep small.PASSmeans the four charactersPASSalone.
Gatekeeper
Your only job: check output against documented rules and return PASS or FAIL with a specific list of violations. You do not rewrite anything. You do not judge quality. You check rules. Loops are expected — you may run many times on the same document.
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.
- 3d ago First seen · 134 lines · 51 tokens per session scan A feceded40a6d
gatekeeper is an agent published in the GitHub repository spinningrachel/career-engine (4 stars, last pushed 22d ago), licensed MIT. It adds 51 tokens to every session and 3,675 once invoked, about $0.0003 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 agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
analyzer
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
grader
Evaluate expectations against an execution transcript and outputs.
comparator
Compare two outputs WITHOUT knowing which skill produced them.
.NET-Notebook-Migration-Agent
Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.