Skip to main content
hc
Submit
Claude Code agents · agents · 16 picks

Best Claude Code agents

Specialized agents for Claude Code — architecture, review, testing, security, and domain expertise.

Curated by @heyclaude-editors Updated 2026-06-19

Specialized agents for Claude Code — architecture, review, testing, security, and domain expertise.

Compared at a glance

The top 5 picks side by side on trust, install, platform support, and disclosed notes — full rationale for each below.

FieldAgent Observability SRE Agent

Community reusable agent prompt for Claude Code analytics and agent platform on-call using official analytics documentation: usage signals, session failure triage, MCP latency patterns, and SRE runbooks for agent hosting teams.

Open dossier 		Agent View Fleet Operator Agent

A reusable agent prompt for running a fleet of Claude Code background sessions from agent view (opened with the agents command). It triages which sessions need input, which are still working, and which have finished, then decides what to dispatch next, which to answer, and which to stop.

Open dossier 		Claude Code Desktop Operations Agent

Community reusable agent prompt for operating Claude Code Desktop using official desktop docs: parallel sessions, agent view, notifications, background jobs, and desktop-specific troubleshooting workflows.

Open dossier 		Microsoft Foundry Claude Code Agent

Community reusable agent prompt for operating Claude Code on Microsoft Foundry using official setup docs: Azure subscriptions, Entra ID auth, deployment names, enterprise network config, and Foundry-specific Claude Code troubleshooting steps.

Open dossier 		Agent Skill QA Reviewer Agent

Source-backed Claude Code subagent prompt for reviewing Agent Skills before adoption or publication, checking SKILL.md scope, descriptions, invocation control, supporting files, tool permissions, helpfulness, safety, and privacy risks against official Claude Code skills guidance.

Open dossier
Trust
Install riskReview firstReview firstReview firstReview firstReview first
Notes Safety ✓ Privacy ✓ Safety ✓ Privacy ✓ Safety ✓ Privacy ✓ Safety ✓ Privacy ✓ Safety ✓ Privacy ✓
Categoryagentsagentsagentsagentsagents
Sourcesource-backedsource-backedsource-backedsource-backedsource-backed
AuthorkiannidevJPette1783kiannidevkiannidevDesel72
Added2026-06-162026-06-052026-06-162026-06-162026-06-08
Platforms
Claude Code
Claude Code
Claude Code
Claude Code
Claude Code
Source repo—————
Safety notes✓Incident commands must not exfiltrate customer prompts into public tickets. Scaling replicas without reviewing tool side effects can amplify destructive MCP calls. Disabling tracing to reduce noise may hide regressions—prefer sampling over full off. Rollback plans should include MCP allowlist and permission settings, not only code.✓This agent prioritizes and routes attention across sessions; it does not bypass each session's own permissions. Background sessions keep running without a terminal attached; ensure each was dispatched with appropriate scope and permissions. Prefer answering or stopping a stuck session over force actions, and review completed sessions before merging their work.✓Parallel desktop sessions on the same branch can race—use worktrees or separate branches. Background jobs inherit directory and permission context from where they were started. Do not approve destructive tool calls in a session you did not start intentionally.✓Client secrets and connection strings must not be committed to repositories. Deployment names and API versions must match Foundry-supported Claude models only. Cross-subscription resources can break policy—keep Foundry resources in approved subscriptions. Azure RBAC changes require admin approval; this agent advises, not applies, policies.✓This agent reviews skill quality and adoption readiness; it does not execute the skill, install plugins, run scripts, or approve production rollout by itself. Flag skills that can perform writes, deployments, destructive actions, account changes, network calls, credential handling, or background automation without explicit user control. Recommend `disable-model-invocation: true`, least-privilege `allowed-tools`, or additional human review when a skill has side effects or could trigger too broadly. Treat supporting files, shell injection blocks, and bundled scripts as executable or instruction-bearing review surfaces, not harmless documentation.
Privacy notes✓Analytics and logs may contain prompts, diffs, and credentials if misconfigured. Recommend redaction before exporting incident timelines externally. Shared dashboards should aggregate metrics without raw user content fields.✓Agent view shows live session activity and working directories; restrict who can view and dispatch sessions. Each session sends its own context to the model provider; dispatch only tasks whose context is acceptable to share. Background sessions may isolate file edits in worktrees; review their diffs before integrating.✓Desktop session titles and diffs may expose proprietary code during screen sharing. Notification previews may leak branch or ticket names—configure OS privacy settings accordingly. Background transcripts follow normal Claude Code data handling for your account type.✓Azure Monitor and Foundry logs may store request metadata; configure retention accordingly. Customer content handling follows Microsoft and Anthropic agreements for Foundry Claude. Support tickets should redact tenant IDs and subscription names when posted publicly.✓Reads local skill instructions and supporting files, which may expose internal workflow names, repository paths, policies, examples, customer data, or credentials accidentally written into prompts. Review output can mention sensitive skill names, tool permissions, file paths, dynamic commands, and risk findings; keep it out of public PR comments unless sanitized. Skills loaded by Claude can place their descriptions or full instructions into model context, so the review should flag secrets and unnecessary confidential details before adoption.
Prerequisites
  • Access to Claude Code analytics or org usage exports for affected teams.
  • Logs from agent hosts, MCP gateways, and background workers when self-hosting SDK workloads.
  • Defined SLOs for session completion time and error budgets for agent tasks.
  • Architecture diagram showing model calls, tool execution, and persistence layers.
  • Claude Code with agent view available (opened with the agents command) and one or more background sessions.
  • A queue of tasks suitable for background dispatch.
  • Knowledge of which tasks are safe to run in parallel and which need close attention.
  • Claude Code Desktop installed per the desktop quickstart for your platform.
  • Multiple concurrent desktop sessions or background jobs requiring coordination.
  • Team agreement on branch ownership and review before merging desktop-generated changes.
  • Optional agent view access for monitoring background sessions.
  • Azure subscription with Microsoft Foundry access and approved Claude deployments.
  • Authentication configured per Foundry docs using Entra ID or approved credential patterns.
  • Corporate network or proxy requirements documented in Claude Code network-config guidance.
  • Enterprise managed settings for Claude Code if rolling out to a large organization.
  • One or more Agent Skill directories with a `SKILL.md` file and any referenced supporting files.
  • Access to the skill's intended user, task boundary, invocation path, and expected output.
  • A clear policy for whether the skill may be model-invoked automatically or must be user-invoked.
  • Permission to inspect tool restrictions, dynamic context commands, scripts, examples, templates, and bundled plugin metadata if present.
Install—————
Config—————
Citations
ClaimUnclaimedUnclaimedUnclaimedUnclaimedUnclaimed
  1. 01
    agents

    Agent Observability SRE Agent

    Operate Claude Code agent platforms using official analytics and on-call runbooks.

    Review firstSource-backedReview firstAdded 3d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Agent Observability SRE Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  2. 02
    agents

    Agent View Fleet Operator Agent

    Keep many background Claude Code sessions productive: spot the ones blocked on a question, queue new work, and retire finished runs.

    Review firstSource-backedReview firstAdded 14d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Agent View Fleet Operator Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  3. 03
    agents

    Claude Code Desktop Operations Agent

    Operate Claude Code Desktop using official parallel session and agent view guidance.

    Review firstSource-backedReview firstAdded 3d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Claude Code Desktop Operations Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  4. 04
    agents

    Microsoft Foundry Claude Code Agent

    Operate Claude Code on Microsoft Foundry using official Azure auth, deployment, and network guidance.

    Review firstSource-backedReview firstAdded 3d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Microsoft Foundry Claude Code Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  5. 05
    agents

    Agent Skill QA Reviewer Agent

    Review Agent Skills for SKILL.md quality, invocation control, helpfulness, safety, privacy, and adoption readiness.

    Review firstSource-backedReview firstAdded 11d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Agent Skill QA Reviewer Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  6. 06
    agents

    Agent Skills Enterprise Librarian Agent

    Reviews an org's SKILL.md library — description and trigger quality, scope and precedence, per-skill allowed-tools, and model-vs-user invocation control.

    Review firstSource-backedReview firstAdded 14d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Agent Skills Enterprise Librarian Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  7. 07
    agents

    AI Agent Cost Governance Analyst Agent

    Govern Claude Code spend using official costs documentation and anomaly triage workflows.

    Review firstSource-backedReview firstAdded 3d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    AI Agent Cost Governance Analyst Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  8. 08
    agents

    Bedrock Claude Code Platform Agent

    Operate Claude Code on Amazon Bedrock using official IAM, region, model ID, and quota guidance.

    Review firstSource-backedReview firstAdded 3d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Bedrock Claude Code Platform Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  9. 09
    agents

    Claude Code Auto Mode Policy Auditor Agent

    Reviews autoMode.environment trust, the hard_deny/soft_deny/allow precedence, and $defaults splicing so the classifier blocks exfiltration without breaking routine internal work.

    Review firstSource-backedReview firstAdded 14d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Claude Code Auto Mode Policy Auditor Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  10. 10
    agents

    Claude Code Enterprise Rollout Agent

    Plans org-wide Claude Code policy: managed-settings deny lists, bypass lockout, managed-MCP allow/deny, fail-closed startup, and audit hooks behind the security approval dialog.

    Review firstSource-backedReview firstAdded 14d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Claude Code Enterprise Rollout Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  11. 11
    agents

    Claude Code Routine Designer Agent

    Design safe recurring Claude Code routines: scoped, idempotent skill-driven workflows with clear triggers, permissions, and reporting.

    Review firstSource-backedReview firstAdded 14d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Claude Code Routine Designer Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  12. 12
    agents

    Claude Code Security Guidance Remediator Agent

    Audit a Claude Code setup against official security guidance and produce a ranked remediation plan: permissions, MCP trust, credentials, and hooks.

    Review firstSource-backedReview firstAdded 4d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Claude Code Security Guidance Remediator Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  13. 13
    agents

    Claude Code Worktree Coordinator Agent

    Coordinate parallel Claude Code sessions across git worktrees: isolation, base branch, file copying, subagent worktrees, and cleanup.

    Review firstSource-backedReview firstAdded 14d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Claude Code Worktree Coordinator Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  14. 14
    agents

    Claude Plugin Marketplace Reviewer Agent

    Vet a Claude Code plugin before install: source trust, the Will-install list, bundled components, context cost, version pinning, and scope.

    Review firstSource-backedReview firstAdded 14d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Claude Plugin Marketplace Reviewer Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  15. 15
    agents

    Claude Slack Delegation Triage Agent

    Triage Slack delegations to Claude Code using official Slack integration documentation.

    Review firstSource-backedReview firstAdded 3d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Claude Slack Delegation Triage Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

  16. 16
    agents

    Computer Use GUI Test Agent

    Validate GUI flows with Claude Code computer use using official enablement and safety docs.

    Review firstSource-backedReview firstAdded 3d ago
    Safety ✓ Privacy ✓
    Why it made the cut

    Computer Use GUI Test Agent is included because it has safety notes present, privacy notes present, source-backed source posture.

    Reach for instead

    If this will touch credentials, local files, or production systems, inspect the upstream source first.

Missing a pick? Propose an edit to this list — every change goes through the same review queue as new entries.

Suggest a pick
Weekly · Sundays

Get the weekly brief

One calm read on Claude workflows. Sundays. No tracking pixels.

Unsubscribe any time. No tracking pixels. No partner blasts.