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 skills add timurgaleev/vibestack --skill bedrock-guardrailsgit clone --depth 1 https://github.com/timurgaleev/vibestackWrote 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/timurgaleev/vibestack/bedrock-guardrails)<a href="https://agentmods.dev/skills/timurgaleev/vibestack/bedrock-guardrails"><img src="https://agentmods.dev/badge/skills/timurgaleev/vibestack/bedrock-guardrails/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/timurgaleev/vibestack/bedrock-guardrails"><img src="https://agentmods.dev/badge/skills/timurgaleev/vibestack/bedrock-guardrails.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 3 findings, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Rogue Agent · line 16 Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
- medium Output Handling · line 285 Output size or generation rate is not bounded. Unbounded output enables denial-of-service through resource exhaustion, log flooding, or context-window stuffing.Fix: Set explicit limits on output length, generation count, and rate. Use max_tokens and truncation to prevent unbounded output.
- medium Excessive Agency · line 525 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00127 | $0.10903 |
| Opus 5 | $0.00063 | $0.05451 |
| Sonnet 5 | $0.00025 | $0.02181 |
| Haiku 4.5 | $0.00013 | $0.01090 |
Grade A, and why
bedrock-guardrails scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
Finding shape: `FAIL HIGH: agent/tools.py:130 model output is passed to subprocess.run without a validator or allowlist.` How it starts
The opening of the file, as written. The whole thing — 530 lines — stays where its author put it; the contents beside it link to each section on GitHub.
When to invoke
Use when: "bedrock guardrails", "review our bedrock setup", "is bedrock locked to eu-central-1", "do we leak PII into prompts", "bedrock IAM policy", "tenant isolation for the knowledge base", "EU data residency for the LLM feature", or when a PR adds a Bedrock invoke and nobody has looked at the surrounding controls yet.
Preamble
eval "$(~/.vibestack/bin/vibe-slug 2>/dev/null)" 2>/dev/null || SLUG="unknown"
_LEARN_FILE="${VIBESTACK_HOME:-$HOME/.vibestack}/projects/${SLUG:-unknown}/learnings.jsonl"
if [ -f "$_LEARN_FILE" ]; then
_LEARN_COUNT=$(wc -l < "$_LEARN_FILE" 2>/dev/null | tr -d ' ')
echo "LEARNINGS: $_LEARN_COUNT entries loaded"
if [ "$_LEARN_COUNT" -gt 5 ] 2>/dev/null; then
~/.vibestack/bin/vibe-learnings-search --limit 5 2>/dev/null || true
fi
else
echo "LEARNINGS: none yet"
fi
{{include lib/snippets/session-host.md}}
{{include lib/snippets/decision-brief.md}}
{{include lib/snippets/working-protocols.md}}
{{include lib/snippets/state-protocols.md}}
User-invocable
When the user types /bedrock-guardrails, run this skill.
Step 1: Locate Bedrock usage and decide the residency regime
Find every place the repo talks to Bedrock. Search infrastructure code and application code separately, because the controls live in different files.
The inventory has to be complete: a call site missed here is skipped by every region, IAM, guardrail, logging and cost check downstream, and shows up in the report as nothing at all. So neither search is truncated, and both cover the whole current invocation surface — InvokeModel, InvokeModelWithResponseStream, Converse, ConverseStream, StartAsyncInvoke, ApplyGuardrail, InvokeGuardrailChecks, RetrieveAndGenerate, RetrieveAndGenerateStream, InvokeAgent, InvokeInlineAgent, InvokeFlow, and the OpenAI-compatible chat/completions and responses paths on the bedrock-runtime endpoint.
grep -rnE 'aws_bedrock|bedrock:|AWS::Bedrock|aws_bedrockagent|bedrock_guardrail' \
--include='*.tf' --include='*.tf.json' --include='*.hcl' --include='*.yaml' --include='*.yml' --include='*.json' --include='*.ts' --include='*.py' . 2>/dev/null \
| grep -vE 'node_modules|\.terraform/|dist/|build/|vendor/'
grep -rnE 'bedrock-(agent-)?runtime|Bedrock(Agent)?Runtime|bedrock_(agent_)?runtime|client\(["'"'"']bedrock|@aws-sdk/client-bedrock|[Ii]nvoke[_]?[Mm]odel([Ww]ith[Rr]esponse[Ss]tream|_with_response_stream)?|[Cc]onverse([_]?[Ss]tream)?[[:space:]]*[(A-Za-z]|[Rr]etrieve[_]?[Aa]nd[_]?[Gg]enerate([_]?[Ss]tream)?|[Ii]nvoke[_]?([Ii]nline[_]?)?[Aa]gent|[Ii]nvoke[_]?[Ff]low|[Ss]tart[_]?[Aa]sync[_]?[Ii]nvoke|[Aa]pply[_]?[Gg]uardrail|[Ii]nvoke[_]?[Gg]uardrail[_]?[Cc]hecks|chat/completions|/v1/responses' \
--include='*.py' --include='*.ts' --include='*.tsx' --include='*.js' --include='*.jsx' --include='*.mjs' --include='*.go' --include='*.java' --include='*.kt' --include='*.rb' --include='*.cs' --include='*.rs' --include='*.php' . 2>/dev/null \
| grep -vE 'node_modules|\.terraform/|dist/|build/|vendor/|_test\.|\.test\.|spec\.'
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.
- 9d ago First seen · 530 lines · 127 tokens per session scan A d455ff5289ae
bedrock-guardrails is a skill published in the GitHub repository timurgaleev/vibestack (6 stars, last pushed 10d ago), licensed MIT. It adds 127 tokens to every session and 10,903 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
bedrock-rag
Build RAG on Amazon Bedrock Knowledge Bases — ingestion, chunking, embeddings, vector stores, Retrieve and RetrieveAndGenerate, citations, and Guardrails contextual grounding. Use when a chatbot must answer from a document corpus with sources.
google-adk-python
Build AI agents with Google Agent Development Kit (ADK) for Python. Use when creating multi-agent systems, tool-using agents, or orchestrating LLM workflows with Google Cloud.
pytorch-training
Train and optimize deep learning models with PyTorch. Use when building neural networks, implementing training loops, or optimizing model performance.
data-engineering
Design and implement data pipelines, ETL processes, and data infrastructure. Use when building data ingestion, transformation, or storage systems.
experiment-tracking
Track ML experiments systematically with MLflow, W&B, or similar tools. Use when running experiments, comparing model versions, or managing reproducibility.
feature-store
Design and manage feature stores for ML systems. Use when building reusable feature pipelines, managing feature versioning, or serving features for inference.