agentleFS
srikanth-manivannansrikanth-manivannan/claude-agent-factory/skills/architecture/tradeoff-analyzer/SKILL.md

tradeoff-analyzer

Skill1 starsChanged 5 months agoMITView on GitHub

Use this to weigh multiple options against weighted criteria and produce a defensible recommendation.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When choosing between ≥2 viable options (libraries, designs, vendors, patterns).

    line 21
  2. Distinctive:

    When a decision needs a written, defensible rationale (e.g.

    line 22
  3. Distinctive:

    When stakeholders disagree and need an explicit, criteria-based comparison.

    line 23
  4. Distinctive:

    When there is only one viable option — there is no trade-off to analyze.

    line 27
  5. Distinctive:

    Score each option per criterion: Use a consistent scale (1–5).

    line 56
  6. Distinctive:

    Run a sensitivity check: Ask: would a reasonable change in the top one or two weights flip the result?

    line 60
  7. Distinctive:

    Never fabricate scores: If evidence for a score is missing, mark it as an assumption (step 8) rather than inventing certainty.

    line 82

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Tradeoff Analyzer

Use this to weigh multiple options against weighted criteria and produce a defensible recommendation.

Tech profile — Technology: any · Language: n/a · Stack: n/a · Toolchain: n/a · Domain: architecture (Pure reasoning skill — applies to any technology decision.)

When to use this skill

  • When choosing between ≥2 viable options (libraries, designs, vendors, patterns).
  • When a decision needs a written, defensible rationale (e.g. feeding an ADR).
  • When stakeholders disagree and need an explicit, criteria-based comparison.

When NOT to use this skill

  • When there is only one viable option — there is no trade-off to analyze.
  • For pure fact lookup (use a research skill).
  • For designing a system from scratch (use architecture/system-design-reviewer to critique a design, or an architecture design skill).

Prerequisites

  • A decision to make and the candidate options (or enough context to enumerate them).
  • The constraints that matter (budget, deadline, team skills, compliance).

Inputs

InputRequiredDescription
decisionyesThe decision to be made, in one sentence
optionsyesThe candidate options (≥2); may be elicited if not given
criterianoEvaluation criteria; derived from constraints if omitted
weightsnoRelative importance per criterion; defaults to equal
constraintsnoHard constraints that disqualify an option outright

Instructions

  1. Frame the decision. Restate it in one sentence and confirm the options (minimum two). If only one option exists, stop and say so.
  2. Apply hard constraints first. Eliminate any option that violates a non-negotiable constraint; record why. Never silently drop an option.
  3. Establish criteria + weights. Derive 4–7 decision criteria from the constraints and goals. Assign each a weight (sum = 100%). State the rationale for the weighting — this is where bias hides, so make it explicit.
  4. Score each option per criterion. Use a consistent scale (1–5). For each score, give a one-line justification grounded in evidence, not vibes.
  5. Compute weighted totals. Multiply scores by weights; sum per option. Present the full matrix (see Output).
  6. Run a sensitivity check. Ask: would a reasonable change in the top one or two weights flip the result? If yes, flag the decision as weight-sensitive.
  7. Recommend. State the winning option, the margin, the key reasons, and the main risk of choosing it. If it is weight-sensitive or close, say so plainly.
  8. Record assumptions. List every assumption made; these are what a reviewer should challenge.

Output

A decision analysis containing:

  • The framed decision and the final option set (with any eliminated options + why).
  • A weighted scoring matrix (options × criteria, scores, weighted totals).
  • A recommendation with margin, top reasons, and primary risk.
  • A sensitivity note (is the result robust?).
  • An assumptions list.

This output is ADR-ready: it drops directly into an ADR's "Decision" and "Consequences" sections.

Constraints & safety

  • Never fabricate scores. If evidence for a score is missing, mark it as an assumption (step 8) rather than inventing certainty.
  • Surface bias. The weighting step must state why each weight was chosen.
  • No false precision. Weighted totals are a decision aid, not a proof; a 0.1 margin is a tie — say so.
  • Conforms to /standards/prompt-engineering.md and /standards/security.md.

Examples

A minimal example is below; full worked examples are in EXAMPLES.md.

Given: Decision = "Which datastore for the events service?", Options = [PostgreSQL, DynamoDB], Criteria = [operational-cost, query-flexibility, team-familiarity, scalability]. Produces: a 2×4 weighted matrix, a recommendation (e.g. PostgreSQL, margin 0.6, weight-sensitive on scalability), and the assumptions behind it.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.