DiegoPrieto23/candidatai
Local job-search app: your AI agent does the reasoning, the app keeps the state, rules and checks. Tailored, verified CVs; assisted applications you submit yourself.
ARCHETYPE
Fort Knox
Deny lists, pre-tool hooks, sandboxing. Nothing touches prod without a signature.
CONTEXT TAX · EVERY TURN
up to ~7.2k tokens
Claude Code loads MCP tools on demand, so the real cost is usually lower.
Heavy · median rig: 2.3k · breakdown
GUARDRAILS
4/5
Blocks destructive commands · Protects secrets · No YOLO mode · Sandbox or ask-first rules · details
Quality 22/100 Guardrails v2
Copy this rig
# review before running: this installs third-party code $ claude mcp add candidatai -- uv run --directory backend cai-mcp $ npx degit DiegoPrieto23/candidatai/.claude ./rig-candidatai # inspect, then merge into .claude/
MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ claude mcp add candidatai -- uv run --directory backend cai-mcp $ npx degit DiegoPrieto23/candidatai/.agents/skills/assess-fit .claude/skills/assess-fit $ npx degit DiegoPrieto23/candidatai/.agents/skills/broaden-search .claude/skills/broaden-search $ npx degit DiegoPrieto23/candidatai/.agents/skills/daily-summary .claude/skills/daily-summary $ npx degit DiegoPrieto23/candidatai/.agents/skills/debrief-interview .claude/skills/debrief-interview $ npx degit DiegoPrieto23/candidatai/.agents/skills/draft-follow-up .claude/skills/draft-follow-up $ npx degit DiegoPrieto23/candidatai/.agents/skills/draft-message .claude/skills/draft-message $ npx degit DiegoPrieto23/candidatai/.agents/skills/expand .claude/skills/expand $ npx degit DiegoPrieto23/candidatai/.agents/skills/generate-cover-letter .claude/skills/generate-cover-letter $ npx degit DiegoPrieto23/candidatai/.agents/skills/generate-cv .claude/skills/generate-cv $ npx degit DiegoPrieto23/candidatai/.agents/skills/handle-requests .claude/skills/handle-requests $ npx degit DiegoPrieto23/candidatai/.agents/skills/import-cv-template .claude/skills/import-cv-template $ npx degit DiegoPrieto23/candidatai/.agents/skills/mock-interview .claude/skills/mock-interview $ npx degit DiegoPrieto23/candidatai/.agents/skills/prepare-application .claude/skills/prepare-application $ npx degit DiegoPrieto23/candidatai/.agents/skills/prepare-interview .claude/skills/prepare-interview $ npx degit DiegoPrieto23/candidatai/.agents/skills/prepare-negotiation .claude/skills/prepare-negotiation $ npx degit DiegoPrieto23/candidatai/.agents/skills/process-jobs .claude/skills/process-jobs $ npx degit DiegoPrieto23/candidatai/.agents/skills/research-company .claude/skills/research-company $ npx degit DiegoPrieto23/candidatai/.agents/skills/review-linkedin .claude/skills/review-linkedin $ npx degit DiegoPrieto23/candidatai/.agents/skills/review-replies .claude/skills/review-replies $ npx degit DiegoPrieto23/candidatai/.agents/skills/review-search .claude/skills/review-search $ npx degit DiegoPrieto23/candidatai/.agents/skills/search-jobs .claude/skills/search-jobs $ npx degit DiegoPrieto23/candidatai/.agents/skills/training-plan .claude/skills/training-plan $ npx degit DiegoPrieto23/candidatai/.agents/skills/update-profile .claude/skills/update-profile
$ curl -fsSL --create-dirs -o .claude/agents/classifier.md https://raw.githubusercontent.com/DiegoPrieto23/candidatai/main/.claude/agents/classifier.md $ curl -fsSL --create-dirs -o .claude/agents/extractor.md https://raw.githubusercontent.com/DiegoPrieto23/candidatai/main/.claude/agents/extractor.md $ curl -fsSL --create-dirs -o .claude/agents/reviewer.md https://raw.githubusercontent.com/DiegoPrieto23/candidatai/main/.claude/agents/reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/verifier.md https://raw.githubusercontent.com/DiegoPrieto23/candidatai/main/.claude/agents/verifier.md $ curl -fsSL --create-dirs -o .claude/agents/writer.md https://raw.githubusercontent.com/DiegoPrieto23/candidatai/main/.claude/agents/writer.md
Merge into .claude/settings.json (project) or ~/.claude/settings.json (user). Hook commands reference scripts in the source repo — copy those too.
{
"permissions": {
"deny": [
"Read(./.env)",
"Read(./.env.*)",
"Read(./data/browser-profile/**)",
"Edit(./data/browser-profile/**)"
],
"ask": [
"Bash(git commit *)",
"Bash(git push *)",
"Bash(uv add *)",
"Bash(pnpm add *)"
]
}
} MCP servers (1)
| server | source | est. tokens |
|---|---|---|
| backend · "candidatai" | pypi | 2.5k |
Skills (23)
assess-fitbroaden-searchdaily-summarydebrief-interviewdraft-follow-updraft-messageexpandgenerate-cover-lettergenerate-cvhandle-requestsimport-cv-templatemock-interviewprepare-applicationprepare-interviewprepare-negotiationprocess-jobsresearch-companyreview-linkedinreview-repliesreview-searchsearch-jobstraining-planupdate-profile
Subagents (5)
| classifier model: haiku | Clasifica UN correo de una empresa (respuesta a una candidatura) de CandidatAI a partir de su reply_id y guarda una propuesta de evento que el usuario aceptará o no. Lo usa el skill review-replies. An |
| extractor model: haiku | Extrae los datos estructurados (JobExtraction) de UNA oferta de CandidatAI a partir de su job_id. Lo usa el skill process-jobs. |
| reviewer model: sonnet | Revisa los cambios de código de CandidatAI contra AGENTS.md y docs/ al cerrar una fase o antes de un commit. Solo lectura. Antes se llamaba revisor. |
| verifier model: sonnet | Verifica de forma independiente que un CV o una carta de presentación de CandidatAI no contienen nada inventado ni exagerado y que cumplen las políticas. Lo usan los skills generate-cv y generate-cove |
| writer model: sonnet | Redacta el contenido (CvContent) de un CV a medida para una oferta aprobada de CandidatAI y lo guarda, lo que renderiza el DOCX/PDF. Lo usa el skill generate-cv. Antes se llamaba redactor. |
Permissions
deny (4)
Read(./.env)
Read(./.env.*)
Read(./data/browser-profile/**)
Edit(./data/browser-profile/**)
ask (4)
Bash(git commit *)
Bash(git push *)
Bash(uv add *)
Bash(pnpm add *)
allow (43)
Bash(uv run pytest *)
Bash(uv run ruff *)
Bash(uv run alembic *)
Bash(uv run cai *)
Bash(uv run --directory backend cai *)
Bash(pnpm test *)
Bash(pnpm lint *)
Bash(pnpm build *)
Bash(pnpm typecheck *)
Bash(git status *)
Bash(git diff *)
Bash(git log *)
mcp__candidatai__get_status
mcp__candidatai__get_schema
mcp__candidatai__list_jobs
mcp__candidatai__get_job
mcp__candidatai__check_job_open
mcp__candidatai__get_search_titles
mcp__candidatai__save_extraction
mcp__candidatai__get_extraction
mcp__candidatai__get_timeline
mcp__candidatai__get_profile
mcp__candidatai__get_context
mcp__candidatai__get_match
mcp__candidatai__get_cv_content
mcp__candidatai__get_cv_brief
mcp__candidatai__list_cv_templates
mcp__candidatai__list_sources
mcp__candidatai__list_mail_sources
mcp__candidatai__list_followups_due
mcp__candidatai__list_upcoming
mcp__candidatai__get_funnel_stats
mcp__candidatai__get_calibration
mcp__candidatai__get_skill_gaps
mcp__candidatai__list_replies
mcp__candidatai__get_reply
mcp__candidatai__save_reply_proposal
mcp__candidatai__save_message_draft
mcp__candidatai__get_job_contacts
mcp__candidatai__get_company_research
mcp__candidatai__get_received_offer
mcp__candidatai__wait_user_action
mcp__candidatai__list_agent_requests
Similar rigs
Sagargupta16/foreman
Claude Code plugin: split a task into subtasks, route each to a predefined agent type on the cheapest model that fits, run them in dependency waves and verify once.
Orchestrator 2.3k tok · · Q0
btclib-org/ai-agents-process
The btclib-org process for coding agents: the btclib-iss, btclib-pr and btclib-table skills, the file they share, and the writer and reviewer agents
Pragmatist 597 tok · · Q0
itshig/AGENTS-template
Drop-in AGENTS.md template for Claude Code, Cursor, Codex, and other AI agents — with .agents/ for rules, personas, and workflows.
Orchestrator 3.0k tok · · Q5
btclib-org/claude-process
The btclib-org process for coding agents: the btclib-iss, btclib-pr and btclib-table skills, the file they share, and the writer and reviewer agents
Pragmatist 524 tok · · Q0