Upwork Job Filter
Evaluate the pasted job against the user's real profile and give an honest, decisive verdict. Never inflate fit — the user's credibility depends on honesty. Read references/profile.md when mapping job requirements to evidence.
1. Extract the facts (ask if missing)
Pull from the posting: title, summary, requirements / "You:" section, engagement type (fixed/hourly, full-time/part-time, one-off/ongoing), budget + estimated hours, "to apply" instructions, and client info (country, reviews, payment verified, $ spent, avg hourly rate paid).
Transferable / adjacent skills (use this in the fit check)
Judge the underlying engineering competency, not the exact tool. Fluency in one tool of a family almost always transfers to its siblings:
- Multi-paradigm dev: FastAPI ↔ Django ↔ Flask; React ↔ Next.js ↔ Preact; pytest ↔ Vitest ↔ Jest.
- Backend/data: PostgreSQL ↔ MySQL ↔ Supabase (Postgres); SQLAlchemy ↔ Prisma ↔ Drizzle.
- AI coding tools: strong fluency with one agentic tool (Codex, Claude Code, Cursor, Copilot) ⇒ credit as fluent in all — same mental model. Treat this as a differentiator, not a checkmark.
Honesty rule for adjacency: it applies within a proven family only. If a job asks for Django and the user has proven a FastAPI/Python backend, count it ✅ (adjacent). But an entirely unproven capability family (browser automation/Playwright, computer vision/OCR, legacy data migration, a domain like CAD) is still a real gap — being good at FastAPI does not make the user a Playwright expert. Say so, and if the core engineering is strong, add "learnable fast" as a note, not as proof.
2. Two-minute filter — a clear NO on any of these → SKIP
- Fit — can the user credibly prove the underlying competency for at least 3 core requirements (exact tool OR a proven adjacent family)? (evidence in
references/profile.md)
- Scope — is it a small/medium fixed task? A full product rebuild, "entire platform", or "multi-year role" → SKIP for now.
- Domain — no unknown jargon (CAD/engineering, fintech, regulated health, etc.) unless the user has proof.
- Ratio — Connects cost vs realistic odds: 18–26 Connects on a low-fit or niche job → SKIP.
3. Price / rate floor
- Effective rate = fixed budget ÷ estimated hours (when scope is given).
- Below ~$15/hr equivalent → SKIP, and show the math (e.g. $96 ÷ 32h = $3/hr).
- Client explicitly wants "lowest rates" → SKIP (bad-signal client).
- Factor Upwork's ~0–15% service fee into the real take-home.
- For a first contract, accept a fair fixed price for credibility, but never exploitation (a $96/32h job is unacceptable).
4. Verdict output format (always explicit)
- VERDICT: APPLY / SKIP / CONDITIONAL
- Fit table: requirement → user evidence → ✅ / ⚠️ / ❌
- 2–3 concrete reasons.
- APPLY / CONDITIONAL → provide a ready-to-paste proposal (respect any required opening phrase, e.g. "SCOPE CONFIRMED.").
- SKIP → one-line why + what to target instead.
5. Proposal drafting (APPLY / CONDITIONAL)
- First 1–2 sentences personalize and prove the job was read (restate the specific need).
- Lead with a real, linked project proof (see
references/profile.md).
- Confirm availability honestly (10–20 h/week; realistic start timing).
- End with 1–2 smart questions.
- Never invent skills, numbers, clients, or results.
Reference
references/profile.md — verified credentials, proof projects, keywords to seek/exclude, red flags.
1---2name: upwork-job-filter3description: Evaluate Upwork (or other freelance) job postings against this user's real, verifiable profile and return an explicit APPLY / SKIP / CONDITIONAL verdict with a fit table and reasons, optionally drafting a proposal. Use when the user pastes a job posting (title, summary, requirements/"You:", client info, budget/rate) and asks whether to apply, asks "fit?", "should I apply?", "what do you think of this job?", or asks you to draft a proposal. Also use to judge whether a Connects spend is justified.4---56# Upwork Job Filter78Evaluate the pasted job against the user's real profile and give an honest, decisive verdict. Never inflate fit — the user's credibility depends on honesty. Read `references/profile.md` when mapping job requirements to evidence.910## 1. Extract the facts (ask if missing)1112Pull from the posting: title, summary, requirements / "You:" section, engagement type (fixed/hourly, full-time/part-time, one-off/ongoing), budget + estimated hours, "to apply" instructions, and client info (country, reviews, payment verified, $ spent, avg hourly rate paid).1314## Transferable / adjacent skills (use this in the fit check)1516Judge the **underlying engineering competency**, not the exact tool. Fluency in one tool of a family almost always transfers to its siblings:1718- **Multi-paradigm dev**: FastAPI ↔ Django ↔ Flask; React ↔ Next.js ↔ Preact; pytest ↔ Vitest ↔ Jest.19- **Backend/data**: PostgreSQL ↔ MySQL ↔ Supabase (Postgres); SQLAlchemy ↔ Prisma ↔ Drizzle.20- **AI coding tools**: strong fluency with one agentic tool (Codex, Claude Code, Cursor, Copilot) ⇒ credit as fluent in all — same mental model. Treat this as a **differentiator**, not a checkmark.2122**Honesty rule for adjacency:** it applies *within a proven family only*. If a job asks for `Django` and the user has proven a FastAPI/Python backend, count it **✅ (adjacent)**. But an **entirely unproven capability family** (browser automation/Playwright, computer vision/OCR, legacy data migration, a domain like CAD) is **still a real gap** — being good at FastAPI does not make the user a Playwright expert. Say so, and if the core engineering is strong, add *"learnable fast"* as a note, not as proof.2324## 2. Two-minute filter — a clear NO on any of these → SKIP25261. **Fit** — can the user credibly prove the **underlying competency** for at least 3 core requirements (exact tool OR a proven adjacent family)? (evidence in `references/profile.md`)272. **Scope** — is it a small/medium fixed task? A full product rebuild, "entire platform", or "multi-year role" → SKIP for now.283. **Domain** — no unknown jargon (CAD/engineering, fintech, regulated health, etc.) unless the user has proof.294. **Ratio** — Connects cost vs realistic odds: 18–26 Connects on a low-fit or niche job → SKIP.3031## 3. Price / rate floor3233- Effective rate = fixed budget ÷ estimated hours (when scope is given).34- Below **~$15/hr** equivalent → SKIP, and show the math (e.g. $96 ÷ 32h = $3/hr).35- Client explicitly wants **"lowest rates"** → SKIP (bad-signal client).36- Factor Upwork's ~0–15% service fee into the real take-home.37- For a **first contract**, accept a *fair* fixed price for credibility, but never exploitation (a $96/32h job is unacceptable).3839## 4. Verdict output format (always explicit)4041- **VERDICT: APPLY / SKIP / CONDITIONAL**42- Fit table: requirement → user evidence → ✅ / ⚠️ / ❌43- 2–3 concrete reasons.44- **APPLY / CONDITIONAL** → provide a ready-to-paste proposal (respect any required opening phrase, e.g. "SCOPE CONFIRMED.").45- **SKIP** → one-line why + what to target instead.4647## 5. Proposal drafting (APPLY / CONDITIONAL)4849- First 1–2 sentences personalize and prove the job was read (restate the specific need).50- Lead with a **real, linked** project proof (see `references/profile.md`).51- Confirm availability honestly (10–20 h/week; realistic start timing).52- End with 1–2 smart questions.53- Never invent skills, numbers, clients, or results.5455## Reference5657- `references/profile.md` — verified credentials, proof projects, keywords to seek/exclude, red flags.
Blue-B-code