agentic-eval

Agentic Evaluation

Use this skill when you are designing or implementing an evaluation loop that lets an agent assess and improve its own outputs through iteration — not when you are running a pre-existing test suite or doing a one-off review with no refinement cycle.

The core pattern is: Generate → Evaluate → Critique → Refine → Output, looping until a convergence condition is met or a max-iteration budget is exhausted.

Use this skill when

• Implementing a self-critique or reflection loop that feeds output quality back into generation.
• Building an evaluator-optimizer pipeline that separates generation from evaluation responsibilities.
• Designing LLM-as-judge scoring to compare or rank multiple candidate outputs.
• Adding rubric-based scoring with weighted dimensions to iterative generation.
• Setting iteration limits, convergence checks, or structured evaluation output contracts.
• The task requires measurable improvement across runs, not just a single-shot best effort.

Do not use this skill when

• You are running an existing test suite to verify code — use verification-before-completion.
• You are diagnosing a specific failure or bug, not evaluating output quality — use systematic-debugging.
• The goal is writing test coverage (unit tests, integration tests) — use test-driven-development.
• You are reviewing a completed artifact once without a refinement loop (a single code review, an editorial pass, a PR check).

Routing boundary

Situation	Use this skill?	Route instead
Designing a reflection loop with a score threshold and max iterations	Yes	—
Implementing LLM-as-judge comparison of two candidate outputs	Yes	—
Running npm test to confirm a fix works	No	verification-before-completion
Tracing why a specific assertion fails	No	systematic-debugging
Writing Jest or pytest test coverage for a module	No	test-driven-development
Reviewing a PR diff once, no iteration	No	review-comment-resolution

Inputs to gather

Required before starting

• The skill or agent behavior to evaluate.
• The target metric: trigger accuracy, refusal rate, or behavioral assertion coverage.

Helpful if present

• Existing evals or trigger-queries files to extend.

First move

1. Identify the skill or behavior to evaluate.
2. Check whether a trigger-queries.json already exists; if so, load it to understand scope.
3. Open the relevant reference file based on the evaluation type.

Navigation

The three evaluation strategy patterns (outcome-based, LLM-as-judge, rubric-based) and full Python examples are in references/patterns.md.

The implementation checklist — criteria, threshold, loop wiring, convergence, logging — is in assets/eval-checklist.md.

For a new implementation, start with the checklist to confirm your setup is complete, then use the patterns reference to choose and adapt an evaluation strategy.

Outputs

• Evaluation loop design with defined criteria, convergence check, and max iteration budget.
• Structured evaluation scores per iteration with input, output, and critique logged.
• Convergence or budget-exhaustion result confirming the loop terminated cleanly.

Workflow

See the body and references for agentic evaluation design and loop steps.

Examples

See references and the skill body for agentic-eval examples.

Reference files

See the references/ directory and linked files in the main content.

Guardrails

• Always set a max_iterations bound (3–5 is a safe default) before wiring up a refinement loop. Unbounded loops stall agents.
• Require structured output (JSON) from the evaluation step so the optimize step has a reliable signal to act on. Free-text critique is fragile.
• Add a convergence check: if the score does not improve between iterations, stop early. Oscillating loops that never converge waste budget.
• Log the full iteration trajectory. Evaluation loops are hard to debug post-hoc without a history of inputs, outputs, scores, and critiques.
• Define evaluation criteria before generating any output. Criteria added mid-loop drift and make scores incomparable across iterations.
• Keep the evaluate step isolated from the generate step. Blending them makes it hard to replace the evaluator or diagnose score instability.
• Handle evaluation parse failures gracefully — if the LLM judge returns malformed JSON, fall back to a safe default (treat as failing) rather than crashing the loop.

Validation

• should trigger: "I want to add a reflection loop to my code-generation agent so it self-critiques and reruns until the score exceeds 0.85"
• should not trigger: "Run the test suite and tell me if the build passes"
• should not trigger: "Why is this specific assertion failing in my TypeScript tests?"

After implementing an evaluation loop, confirm:

• max_iterations is set and respected by the loop
• Evaluate step returns structured output (JSON or equivalent)
• Convergence check exits early when score does not improve
• All iterations are logged with input, output, score, and critique
• Parse-failure fallback is present on the evaluate step
• Criteria are defined before any generation begins

Examples

• "Add a self-critique loop to my report-generation agent that retries up to three times if the rubric score is below 0.8."
• "Implement an evaluator-optimizer where a separate LLM judge scores code clarity and the generator rewrites until it passes."
• "Build a rubric-based evaluator with accuracy, completeness, and style dimensions that returns a weighted score as JSON."

Reference files

• references/patterns.md — The three evaluation strategy patterns (outcome-based, LLM-as-judge, rubric-based) with annotated Python examples and a best-practices table.
• assets/eval-checklist.md — Implementation checklist: setup, loop wiring, convergence, logging, and safety items to confirm before shipping.