agents-harden

Use when preparing your agent for production — IAM scoping, inbound auth (JWT, SigV4), secrets management, cold start optimization, session lifecycle, rate limiting, input validation, and quota guidance. Triggers on: "production checklist", "harden agent", "production ready", "secure agent", "inbound auth", "going live", "cold start optimization", "session lifecycle", "StopRuntimeSession", "quota", "throttling", "maxVms", "rate limit", "security audit of outbound API calls", "gateway target audit for production", "restrict who can call", "lock down endpoint", "only our app can call". Not for Cedar tool-restriction policies — use agents-connect. Not for quality measurement — use agents-optimize. Not for outbound credential storage or API key wiring — use agents-connect. Not for A2A agent-to-agent auth — use agents-build. Cold start observation and diagnosis (not optimization) routes to agents-debug.

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

77%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A thorough, highly actionable hardening guide with concrete commands, policies, and code, plus clear sequenced workflows with validation checkpoints for the riskiest operations. Its main weaknesses are length/prose that could be tightened and a large body of reference-grade detail kept inline rather than factored into deeper reference files.

Suggestions

Trim inferable explanatory prose (e.g., the cold-start time breakdown narrative and the CodeZip-vs-Container historical aside) to lift conciseness toward the top anchor.

Factor the longer inline sections — JWT auth deep-dive, session lifecycle, and cold-start optimization — into one-level-deep reference files linked from a leaner SKILL.md overview to improve progressive disclosure.

Tighten the governing 'Step 2: Run through the checklist' with an explicit per-category validate→report loop so the audit workflow matches the rigor of the maxVms and JWT debugging sequences.

Dimension	Reasoning	Score
Conciseness	Largely actionable but padded with explanatory prose a competent Claude could infer — e.g., the "Where cold start time actually goes" narrative breakdown and "The skill previously recommended CodeZip over Container when possible. That's an oversimplification" — so it is mostly efficient but could be tightened.	2 / 3
Actionability	Copy-paste-ready throughout: executable bash (`agentcore status --json \| jq -r '.runtimes[0].executionRoleArn'`), complete IAM policy JSON, and runnable Python using `@requires_api_key`/`@requires_access_token` decorators and `StopRuntimeSession`/`add_async_task` calls.	3 / 3
Workflow Clarity	Clear top-level sequence (Step 0 verify CLI version → Step 1 read project → Step 2 run checklist) plus explicit ordered validation/feedback loops for the fragile paths, such as the four-step maxVms diagnosis and the ordered JWT 403 debugging walkthrough.	3 / 3
Progressive Disclosure	The single reference file `references/limits.md` is real and clearly signaled inline, and cross-skill links are well marked, but a large amount of reference-grade detail (IAM policies, JWT deep-dive, cold start, session lifecycle) lives inline rather than split into separate one-level-deep files.	2 / 3
	Total	10 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, specific description that clearly states both purpose and triggers, with rich natural-language trigger terms and thorough negative routing to neighboring skills. No first/second-person voice issues; not verbose.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete production-hardening actions — "IAM scoping, inbound auth (JWT, SigV4), secrets management, cold start optimization, session lifecycle, rate limiting, input validation, and quota guidance" — matching the top anchor for naming several specific concrete actions.	3 / 3
Completeness	Explicitly answers both what (security, reliability, performance hardening) and when, opening with "Use when preparing your agent for production" and an explicit "Triggers on:" list, satisfying the top anchor.	3 / 3
Trigger Term Quality	Broad natural trigger coverage a user would actually say — "production checklist", "harden agent", "production ready", "going live", "rate limit", "restrict who can call", "lock down endpoint", "only our app can call" — alongside relevant technical terms.	3 / 3
Distinctiveness Conflict Risk	Extensive negative routing ("Not for Cedar tool-restriction policies — use agents-connect", "Not for quality measurement — use agents-optimize", "Not for A2A agent-to-agent auth — use agents-build") carves a clear niche and steers away from sibling skills, minimizing wrong-skill triggering.	3 / 3
	Total	12 / 12 Passed

Validation

81%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (705 lines); consider splitting into references/ and linking	Warning
relative_links	Relative link issues: 5 suspicious	Warning
referenced_paths_exist	Referenced path issues: 1 missing	Warning

	Total	13 / 16 Passed

Repository: aws/agent-toolkit-for-aws
Path: plugins/aws-agents/skills/agents-harden/SKILL.md
Commit: 222ce56

Reviewed: 3 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.