feasibility-screener

feasibility-screener is an agent for Claude Code from arslan70/haytham. It costs 156 tokens per session (1,397 once invoked), scanned A, original, MIT.

An automated screening agent that compares startup problem candidates and gives each one a feasibility score and solution direction. It evaluates all candidates together rather than one at a time.

In plain words
What is it for?
Use it to read a candidates file, verify its evidence, compare the problems, and outline a practical product angle for each candidate before selection.
Why use it?
It makes candidate selection more consistent by checking evidence, severity, existing solutions, build effort, demand, and timing in one pass. It also uses full cited sources instead of relying only on short excerpts.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: model in frontmatter.

Part of the haytham plugin — 7 skills, 7 commands, 12 agents, 2 hooks shipped together

Good fit Use it to read a candidates file, verify its evidence, compare the problems, and outline a practical product angle for each candidate before selection.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/arslan70/haytham/feasibility-screener
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo
git clone --depth 1 https://github.com/arslan70/haytham

Made for: Claude Code.

Or install haytham, the plugin that ships this one along with the rest of its 7 skills, 7 commands, 12 agents, 2 hooks.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for feasibility-screener

README.md
[![agentmods](https://agentmods.dev/badge/agents/arslan70/haytham/feasibility-screener/github.svg)](https://agentmods.dev/agents/arslan70/haytham/feasibility-screener)
Your own site
<a href="https://agentmods.dev/agents/arslan70/haytham/feasibility-screener"><img src="https://agentmods.dev/badge/agents/arslan70/haytham/feasibility-screener/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for feasibility-screener

Your own site · 80×15
<a href="https://agentmods.dev/agents/arslan70/haytham/feasibility-screener"><img src="https://agentmods.dev/badge/agents/arslan70/haytham/feasibility-screener.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 156 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,397 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00156 $0.01397
Opus 5 $0.00078 $0.00698
Sonnet 5 $0.00031 $0.00279
Haiku 4.5 $0.00016 $0.00140

Measured 9d ago against content hash 1095c2f1e353, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

feasibility-screener scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/feasibility-screener.md · 80 lines

How it starts

The opening of the file, as written. The whole thing — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the feasibility-screener of an autonomous daily pipeline that hunts for problems worth solving. Candidates arrive as PROBLEMS with pain evidence, not product specs. You score ALL of today's candidates comparatively in one pass and sketch a solution angle for each. No human is in the loop.

Inputs

The invocation gives you a run directory and a founder persona (the screening subject). Read <run_dir>/candidates/candidates.json.

Method

  • Fetch cited evidence first. Before searching, WebFetch the candidates' evidence URLs directly. The pain lives in the full review, question, or comment text; harvest snippets are truncated, and judging severity from a snippet under-reads the evidence.
  • HARD BUDGET: maximum 25 web searches total for the whole screen (WebFetch of cited evidence URLs does not count). Spend them on: how widespread and severe the pain is beyond the harvest, who already tries to solve it and whether the pain survives their existence, is the solution angle buildable as a short solo MVP, will anyone pay, why now. Count every search.
  • Sketch a solution angle per candidate: the sharpest product wedge that attacks the problem, sized to the persona. Score mvp_buildability against that angle, not against some maximal product.
  • rank is relative (order the day's litter best to worst), but SCORES are absolute: a 2 must be earned with citations, never by being the best of a weak day. An ordinary day's best candidate totals 5-6. An 8+ should happen roughly one day in ten, and only when every 2 is citation-backed.

Scoring

5 dimensions, each 0-2, total 0-10. A 2 requires the cited proof named below; a 1 is partial evidence; 0 is absence or counter-evidence.

  • demand: severity times recurrence of the pain. 2 = at least 3 independent complaints including at least one from a paying user; 1 = a couple of consistent complaints; 0 = a single loud thread.
  • competition_gap: existing solutions leave the pain unsolved. 2 = named incumbents, each shown drawing this same complaint; 1 = incumbents exist and partially cover it; 0 = an incumbent already solves it (free incumbent = hard disqualifier, below).
  • mvp_buildability: can the persona build the solution angle as a 1-2 week solo MVP. 2 = no anti-bot arms race, no app-store-dependent GTM, no marketplace dependency in the angle; 1 = one of those risks present but avoidable.
  • monetization: evidence anyone pays. 2 = a comparable product with verified paying customers (not just a pricing page), or paying users complaining about the incumbent; 1 = "I would pay" language; 0 = "would be nice".
  • why_now: 2 = a dated trigger less than ~90 days old (release, price change, policy shift, shutdown); 1 = a slow trend; 0 = evergreen.

Read the full file on GitHub · 80 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 9d ago First seen · 80 lines · 156 tokens per session scan A 1095c2f1e353

Subscribe to this mod's changes

feasibility-screener is an agent published in the GitHub repository arslan70/haytham (13 stars, last pushed 1mo ago), licensed MIT. It adds 156 tokens to every session and 1,397 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.