PROMPT SKETCH
Generate executable candidates in an isolated runtime, evaluate every selected candidate, and retain execution and fitness provenance for the archive.Evaluator-guided program search
Evolve executable candidates.
JOIN POINT
executable_candidate_generationRECORD STATUSCandidate · v0.1.001
UML class view
The proposed contract drawn as participant roles, received calls, and numbered dependencies.
02
Object interactions
The same roles and calls as a class collaboration and a numbered message sequence.
03
Pointcut
Where the mechanism applies. All four dimensions belong to the proposed match, not just its vocabulary.
- Context
- Harness or harness-adjacent behavior is represented as executable programs.
- Trigger
- Candidate programs are generated, mutated, or selected.
- Structure
- Generator, executor, fitness evaluator, archive, and next-generation policy.
- Effect
- A next generation is selected from measured executable candidates.
04
Protected contract
The obligation that clients, operators, and auditors should be able to rely on.
- Requires
- Candidate runtime and evaluation suite are isolated and available.
- Ensures
- Every selected candidate has execution and fitness provenance.
- Invariant
- Evaluation budget and security boundaries hold across all candidates.
- Violation
- Crash, timeout, unsafe code, or invalid fitness removes or quarantines candidate.
Negative control
Prompt-level variation with no executable search is not E07.
05
Variation & lifecycle
The hot spot is admitted only where the protected semantics survive.
Permitted variation
AlphaEvolve-like search, evolutionary archive, program mutation, and fitness gates.
Test obligation
Test execution, unsafe candidate, timeout, fitness error, and generation update.
Evolution question
Did program search become part of harness engineering?
06
Prompt + code design
An illustrative prompt and a small design sketch for the proposed contract. Names and APIs are invented; this is not a repository excerpt.
CODE DESIGN / PYTHON
def search(task, evaluator, archive):
candidate = generator.propose(task, archive)
with sandbox.run(candidate) as run:
result = run.output()
fitness = evaluator.score(result, task)
archive.add(CandidateRecord(candidate, result, fitness))
return archive.best()Illustrative design sketch. Selection is only as trustworthy as the isolation and provenance of the run. This example is a teaching aid, not implementation evidence or a conformance claim.
07
Detection & matching
The frozen rule that retrieves this record, the canonical source exemplar, and the control that should stay unmatched.
- Join point
executable_candidate_generation- Source locator
- See the source record.
- Lexical cue
(?i)\b(?:program[_-]?search|evolve\w*\s+programs?|AlphaEvolve|fitness[_-]?evaluation)\bFrozen expression, v0.1.0. A match is a retrieval cue, not a confirmed implementation.- Aliases
- programsearch, fitnessevaluation, evolveprogram, programevolution
- Canonical exemplar
- Darwin-Agent/HarnessX
recipe/gaia_evolver/run_meta.py:437Support: tests/unit/test_gaia_gating.py:1 - Negative control
- Prompt-level variation with no executable search is not E07.
Scope-anchored review. pohe-agent analyze <repo> --pattern E07 anchors the cue to its enclosing declaration and checks a separate operational signature for all four pointcut dimensions. Missing roles and negative-control risks remain visible; a structural match is not E2/E3. Calibration report ↗
08
Repository files
Up to five public files retrieved by the frozen cue and pinned to the audited commit. A match is a place to inspect, not a confirmed implementation.
09
Evidence & provenance
Each layer answers a different question. Do not substitute retrieval for review.
Source-located exemplar
A concrete source anchor is available for this proposed record. It is one exemplar, not evidence that all repositories—or every contract field—conform.
Review boundary: this candidate and its specification come from an AI-assisted study. Independent human validation remains open.
ANY-EVIDENCE FAMILY COVERAGE3/36
8.3% of repository families in the frozen panel have at least one current cue, historical cue, or canonical source layer. This is not confirmed prevalence.
Current lexical-cue repositories: 2 / 41Canonical source anchorDarwin-Agent/HarnessX
ccb411691343recipe/gaia_evolver/run_meta.py:437Supporting surface
tests/unit/test_gaia_gating.py:1Evaluate candidate harnesses and evolve the next round.
Exploratory retrieval cue Regex / v0.1.0
(?i)\b(?:program[_-]?search|evolve\w*\s+programs?|AlphaEvolve|fitness[_-]?evaluation)\bThis expression is a discovery instrument only. A match is a retrieval cue—not a confirmed implementation, pointcut match, or contract proof.