{"id":50760516,"url":"https://github.com/ek33450505/cast-agents","last_synced_at":"2026-06-11T09:30:53.866Z","repository":{"id":357147016,"uuid":"1199935469","full_name":"ek33450505/cast-agents","owner":"ek33450505","description":"CAST specialist agent definitions — 23 production-grade Claude Code agents (commit, debug, review, plan, test, research, and more)","archived":false,"fork":false,"pushed_at":"2026-05-25T22:09:48.000Z","size":175,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-26T00:23:31.069Z","etag":null,"topics":["agent-framework","agents","ai-agents","anthropic","cast","claude-code","cli","developer-tools","open-source","specialists"],"latest_commit_sha":null,"homepage":null,"language":"Shell","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/ek33450505.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":"CODE_OF_CONDUCT.md","threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":"SECURITY.md","support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-04-02T21:34:09.000Z","updated_at":"2026-05-25T22:09:52.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/ek33450505/cast-agents","commit_stats":null,"previous_names":["ek33450505/cast-agents"],"tags_count":4,"template":false,"template_full_name":null,"purl":"pkg:github/ek33450505/cast-agents","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ek33450505%2Fcast-agents","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ek33450505%2Fcast-agents/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ek33450505%2Fcast-agents/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ek33450505%2Fcast-agents/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ek33450505","download_url":"https://codeload.github.com/ek33450505/cast-agents/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ek33450505%2Fcast-agents/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":34192870,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-06-11T02:00:06.485Z","response_time":57,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agent-framework","agents","ai-agents","anthropic","cast","claude-code","cli","developer-tools","open-source","specialists"],"created_at":"2026-06-11T09:30:53.723Z","updated_at":"2026-06-11T09:30:53.857Z","avatar_url":"https://github.com/ek33450505.png","language":"Shell","funding_links":[],"categories":[],"sub_categories":[],"readme":"# cast-agents\n\n[![CI](https://github.com/ek33450505/cast-agents/actions/workflows/ci.yml/badge.svg)](https://github.com/ek33450505/cast-agents/actions/workflows/ci.yml)\n![version](https://img.shields.io/badge/version-0.5.1-blue)\n![license](https://img.shields.io/badge/license-MIT-green)\n![platform](https://img.shields.io/badge/platform-macOS%20%7C%20Linux-lightgrey)\n![agents](https://img.shields.io/badge/agents-23-purple)\n\n23 specialist Claude Code agent definitions. Drop them into `~/.claude/agents/` and they are immediately available in any Claude Code session. Commit, review, debug, plan, research, test — each task goes to the right expert.\n\n## What you get\n\nEvery agent is a markdown file with a YAML frontmatter block that Claude Code reads as a named sub-agent. Dispatch any agent by name from within a session:\n\n\u003e \"Use the code-reviewer agent to review my changes\"\n\u003e \"Dispatch the debugger agent to fix this TypeScript error\"\n\u003e \"Use the researcher agent to analyze this architectural question\"\n\nAgents handle one concern well, use the right model for the job (haiku for speed, sonnet for reasoning), and always end with a structured `Status:` block so orchestrators can route their output reliably.\n\n## Install\n\n### Homebrew\n\n```bash\nbrew tap ek33450505/cast-agents\nbrew install cast-agents\ncast-agents install\n```\n\n### Manual\n\n```bash\ngit clone --recurse-submodules https://github.com/ek33450505/cast-agents.git\ncd cast-agents\nbash install.sh\n```\n\n## Agent roster\n\n### General-purpose agents\n\nWork standalone with any Claude Code setup — no CAST framework required.\n\n| Agent | Model | Effort | What it does |\n|---|---|---|---|\n| `api-contract` | sonnet | medium | API contract guardian. Detects breaking changes in REST endpoints, compares route signatures and response shapes, generates OpenAPI-style diffs |\n| `bash-specialist` | haiku | low | Complex shell scripts, debugging Bash, BATS tests |\n| `code-reviewer` | haiku | low | Immediate post-change code review |\n| `code-writer` | sonnet | high | Feature implementation, refactors, bug fixes |\n| `commit` | haiku | low | Semantic git commit message generation |\n| `debugger` | sonnet | high | Error triage, root cause analysis, fixes |\n| `dep-auditor` | haiku | low | Dependency auditor. Reviews package changes for transitive risk, known CVEs, version compatibility, and license concerns |\n| `devops` | haiku | low | CI/CD, Docker, shell automation, deployments |\n| `docs` | haiku | low | Documentation, changelogs, README files |\n| `eval-writer` | sonnet | medium | Eval and benchmark fixture author for Claude API and CAST agent prompts. Generates regression fixtures that catch prompt-level behavior drift |\n| `frontend-qa` | haiku | low | UI component review and accessibility checks |\n| `merge` | haiku | low | Git merge, rebase, and conflict resolution — merges branches, resolves conflicts, prunes dead worktrees |\n| `migration-reviewer` | opus | high | Database schema change reviewer. Analyzes migration files for safety, generates rollback plans, validates ordering, and checks for data-loss risks |\n| `perf-sentinel` | sonnet | medium | Performance regression detector. Runs benchmarks, interprets results in context of recent changes, and suggests git bisect targets |\n| `planner` | sonnet | high | Write structured plans with Agent Dispatch Manifests |\n| `pr-reviewer` | sonnet | medium | Holistic pull-request reviewer. Reads the full diff, commit history, and linked issues at PR-open time |\n| `push` | haiku | low | Safe git push with pre-push validation |\n| `release-notes` | haiku | low | Release notes generator. Creates structured changelogs from git commits, resolved issues, and breaking changes between two refs |\n| `researcher` | sonnet | high | Deep investigation, multi-step analysis |\n| `security` | sonnet | medium | Security audits, vulnerability assessment |\n| `test-runner` | haiku | low | Run test suites, report failures |\n| `test-writer` | haiku | low | Write tests for new features and edge cases |\n\n### CAST-enhanced agents\n\nWork standalone, but produce richer output when the CAST observability database (`cast.db`) is present.\n\n| Agent | Model | Effort | What it does |\n|---|---|---|---|\n| `morning-briefing` | haiku | low | Daily briefing: git activity, plans, session summary |\n\n## The Status block contract\n\nEvery agent ends its response with a structured Status block:\n\n```\nStatus: DONE\nSummary: Reviewed 3 files, found 2 issues, both addressed.\n```\n\nValid values:\n\n| Status | Meaning |\n|---|---|\n| `DONE` | Task completed successfully |\n| `DONE_WITH_CONCERNS` | Completed, but human attention needed on something |\n| `BLOCKED` | Cannot proceed — missing input, permission, or context |\n| `NEEDS_CONTEXT` | Stopped to ask a clarifying question before continuing |\n\nThis contract is what makes automated orchestration reliable. The orchestrator reads the Status line to decide whether to continue, retry, or stop and wait for human intervention — no natural language parsing required.\n\nExample `DONE_WITH_CONCERNS`:\n```\nStatus: DONE_WITH_CONCERNS\nSummary: Tests written for 5 functions.\nConcerns: coverage/auth.ts is at 61% — below the 80% threshold. Consider adding edge case tests for token expiry.\n```\n\n## Model selection rationale\n\nHaiku agents handle high-frequency, low-complexity tasks — commit messages, code review, push validation, running tests, doc writing, shell scripting, devops, morning briefings, dep auditing, release notes, and test writing. Sonnet agents handle reasoning-heavy work — writing code, debugging, planning, research, security review, API contracts, eval writing, PR review, and performance analysis. Opus is reserved for migration-reviewer, where the cost of a missed data-loss risk outweighs the model price difference. The 13 haiku / 9 sonnet / 1 opus split and the cost differences between models makes this worth understanding before you dispatch everything to sonnet.\n\nRule of thumb: if the task requires reading and deciding (not reading and writing), start with haiku.\n\n## Usage examples\n\n```\n# After implementing a feature:\n\"Use the commit agent to commit my staged changes\"\n\n# After writing code:\n\"Use the code-reviewer agent to review the changes in src/api/\"\n\n# When something breaks:\n\"Dispatch the debugger agent — getting TypeError: Cannot read property 'id' of undefined in line 42\"\n\n# Before a complex task:\n\"Use the planner agent to write a plan for refactoring the auth module\"\n\n# Research question:\n\"Use the researcher agent to analyze tradeoffs between JWT and session-based auth for our stack\"\n```\n\n## Works with CAST\n\nThese agents are extracted from [CAST](https://github.com/ek33450505/claude-agent-team) — a full multi-agent framework built on Claude Code with orchestration, hooks, observability (cast.db), and automated plan execution. If you want the complete framework — including the `planner` agent working at full capacity with ADM manifests, plus CI hooks and a dashboard — install CAST.\n\ncast-agents is for teams and individuals who want the specialized agents without the full framework.\n\n## CAST Ecosystem\n\n\u003e Auto-synced from [claude-agent-team/docs/ecosystem.md](https://github.com/ek33450505/claude-agent-team/blob/main/docs/ecosystem.md). Run `~/Projects/personal/claude-agent-team/scripts/sync-ecosystem-readme.sh` to refresh.\n\n\u003c!-- ECOSYSTEM_START --\u003e\n| Repo | Description | Latest | Install |\n|---|---|---|---|\n| [cast-hooks](https://github.com/ek33450505/cast-hooks) | 13 auditable hook scripts — observability, safety guards, quality gates. SessionStart, PreToolUse, PostToolUse, PostCompact. | ![](https://img.shields.io/github/v/release/ek33450505/cast-hooks?style=flat-square) | `brew tap ek33450505/cast-hooks \u0026\u0026 brew install cast-hooks` |\n| [cast-agents](https://github.com/ek33450505/cast-agents) | 23 specialist agents — commit, debug, review, plan, test, research, and more. Agent definitions with YAML frontmatter. v7-synced. | ![](https://img.shields.io/github/v/release/ek33450505/cast-agents?style=flat-square) | `brew tap ek33450505/cast-agents \u0026\u0026 brew install cast-agents` |\n| [cast-memory](https://github.com/ek33450505/cast-memory) | Persistent agent memory with FTS5 search, relevance scoring, shared pool, semantic embeddings. Per-agent knowledge accumulation. | ![](https://img.shields.io/github/v/release/ek33450505/cast-memory?style=flat-square) | `brew tap ek33450505/cast-memory \u0026\u0026 brew install cast-memory` |\n| [cast-routines](https://github.com/ek33450505/cast-routines) | Scheduled autonomous Claude Code routines via YAML + cron. Daily briefings, inbox triage, release celebration, weekly cost reports. | ![](https://img.shields.io/github/v/release/ek33450505/cast-routines?style=flat-square) | `brew tap ek33450505/cast-routines \u0026\u0026 brew install cast-routines` |\n| [cast-parallel](https://github.com/ek33450505/cast-parallel) | Parallel agent execution across worktree sessions. Agent Dispatch Manifest (ADM) support. | ![](https://img.shields.io/github/v/release/ek33450505/cast-parallel?style=flat-square) | `brew tap ek33450505/cast-parallel \u0026\u0026 brew install cast-parallel` |\n| [cast-observe](https://github.com/ek33450505/cast-observe) | Session-level observability — cost tracking, agent run history, token spend, event sourcing. Feeds cast.db. | ![](https://img.shields.io/github/v/release/ek33450505/cast-observe?style=flat-square) | `brew tap ek33450505/cast-observe \u0026\u0026 brew install cast-observe` |\n| [cast-security](https://github.com/ek33450505/cast-security) | Security hooks and audit trails. PII redaction, parry-guard integration, compliance logging. | ![](https://img.shields.io/github/v/release/ek33450505/cast-security?style=flat-square) | `brew tap ek33450505/cast-security \u0026\u0026 brew install cast-security` |\n| [cast-doctor](https://github.com/ek33450505/cast-doctor) | Read-only health check for any Claude Code install. Validates hooks, MCP servers, agent frontmatter, cast.db schema, stale memories. | ![](https://img.shields.io/github/v/release/ek33450505/cast-doctor?style=flat-square) | `brew tap ek33450505/cast-doctor \u0026\u0026 brew install cast-doctor` |\n| [cast-time](https://github.com/ek33450505/cast-time) | Gives Claude Code a clock — injects local time, timezone, and a semantic time-of-day bucket at every SessionStart. | ![](https://img.shields.io/github/v/release/ek33450505/cast-time?style=flat-square) | `brew tap ek33450505/cast-time \u0026\u0026 brew install cast-time` |\n| [cast-dash](https://github.com/ek33450505/cast-dash) | Terminal UI dashboard for live swarm monitoring. 4-panel real-time display (Textual framework). | ![](https://img.shields.io/github/v/release/ek33450505/cast-dash?style=flat-square) | `brew tap ek33450505/cast-dash \u0026\u0026 brew install cast-dash` |\n| [cast-claudes_journal](https://github.com/ek33450505/cast-claudes_journal) | Session continuity — Claude's Journal auto-injects prior-day context via SessionStart hook. Obsidian vault sync. | ![](https://img.shields.io/github/v/release/ek33450505/cast-claudes_journal?style=flat-square) | `brew tap ek33450505/homebrew-claudes-journal \u0026\u0026 brew install claudes-journal` |\n| [cast-website](https://github.com/ek33450505/cast-website) | castframework.dev — marketing site and docs portal for the CAST ecosystem. | ![](https://img.shields.io/github/v/release/ek33450505/cast-website?style=flat-square) | — |\n| [cast-desktop](https://github.com/ek33450505/cast-desktop) | Tauri 2 native app — embedded PTY terminal, command palette, 11 dashboard views. | ![](https://img.shields.io/github/v/release/ek33450505/cast-desktop?style=flat-square) | `brew tap ek33450505/homebrew-cast-desktop \u0026\u0026 brew install cast-desktop` |\n\u003c!-- ECOSYSTEM_END --\u003e\n\n## License\n\nMIT — see [LICENSE](LICENSE)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fek33450505%2Fcast-agents","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fek33450505%2Fcast-agents","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fek33450505%2Fcast-agents/lists"}