aws-samples/sample-pellier-agentic-search-apg is an educational reference application for an AI shopping concierge backed by Amazon Aurora PostgreSQL and Amazon Bedrock AgentCore. Its fictional retail storefront turns natural-language shopping requests into catalog searches, specialist-agent decisions, tool calls, and recorded business state, while a companion engineering interface exposes how those operations work. The catalogue add-ons support using or understanding this reference application.
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 aws-samples/sample-pellier-agentic-search-apg --skill evidencegit clone --depth 1 https://github.com/aws-samples/sample-pellier-agentic-search-apgWrote 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/aws-samples/sample-pellier-agentic-search-apg/evidence)<a href="https://agentmods.dev/skills/aws-samples/sample-pellier-agentic-search-apg/evidence"><img src="https://agentmods.dev/badge/skills/aws-samples/sample-pellier-agentic-search-apg/evidence/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/aws-samples/sample-pellier-agentic-search-apg/evidence"><img src="https://agentmods.dev/badge/skills/aws-samples/sample-pellier-agentic-search-apg/evidence.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 | $0.00038 | $0.00344 |
| Opus 5.5 | $0.00015 | $0.00138 |
| Sonnet 5.5 | $0.00008 | $0.00069 |
| Haiku 4.5 | $0.00004 | $0.00034 |
Grade A, and why
evidence 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 4d 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
The Proof Counter
When to apply
- "Why this?", "how do you know?", "what did you use?", "show me proof", or trace/debug asks.
- Questions about memory, prior orders, source grounding, Gateway, policy, audit receipts, or whether a tool call was allowed.
Voice and proof rules
- Answer from evidence first: named tool, memory source, audit row, or retrieved product attribute.
- Keep proof concise enough for a shopper, but precise enough for an operator to inspect.
- Separate shopper-facing meaning from operator-facing evidence in one clean sentence when both matter.
- Prefer concrete nouns: product name, tool name, caller rail, row id, policy decision, memory table.
Tool discipline
- Use the persona context and AgentCore Memory for preference proof; no tool reads preferences.
- Use
get_ordersandget_ticketsfor what the records show about the signed-in shopper. - Use retrieval tools only when the shopper asks why a specific recommendation fits and no receipt or memory fact is already enough.
Guardrails
- Do not claim a tool ran unless a receipt, trace event, or tool result confirms it.
- Do not expose secrets, raw JWTs, or infrastructure identifiers that are not already surfaced by the app.
- When no receipt exists, say that no ALLOW receipt was found; do not invent a denial reason.
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.
- 4d ago First seen · 34 lines · 38 tokens per session scan A 591d7fa47eb8
evidence is a skill published in the GitHub repository aws-samples/sample-pellier-agentic-search-apg (12 stars, last pushed yesterday), licensed MIT. It adds 38 tokens to every session and 344 once invoked, about $0.0002 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 2026-11-02.
evidence compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| evidenceaws-samples/sample-pellier-agentic-search-apg | Skill | 12 | yesterday | MIT | A | 38 | Claude Code, Codex |
| xerjxerj-org/xerj | Skill | 19k | today | Apache-2.0 | A | 38 | Claude Code, Codex |
| code-interpreteraws-samples/sample-strands-agent-with-agentcore | Skill | 197 | 5d ago | MIT | A | 25 | Claude Code, Codex |
| code-agentaws-samples/sample-strands-agent-with-agentcore | Skill | 197 | 5d ago | MIT | A | 47 | Claude Code, Codex |
| excalidrawaws-samples/sample-strands-agent-with-agentcore | Skill | 197 | 5d ago | MIT | A | 19 | Claude Code, Codex |
| excel-spreadsheetsaws-samples/sample-strands-agent-with-agentcore | Skill | 197 | 5d ago | MIT | A | 14 | Claude Code, Codex |
| word-documentsaws-samples/sample-strands-agent-with-agentcore | Skill | 197 | 5d ago | MIT | A | 13 | Claude Code, Codex |