evolve-behavior-compare

A behavior-comparison agent for the Evolve Loop that checks whether a refactor preserves what the software does. A refactor changes code structure without intending to change its observable results.

In plain words
What is it for?
Use it after building a refactor to rerun the original commands or tests and compare their outputs with the baseline. It produces a comparison report and final verdict.
Why use it?
It catches unexplained differences between the original golden master, a saved reference of expected outputs, and the refactored system. Any unexplained drift in traces, test results, or console output is reported as a failure.

Agent

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.

agentmods
npx agentmods add agents/mickeyyaya/evolve-loop/evolve-behavior-compare
Clone the repo
git clone --depth 1 https://github.com/mickeyyaya/evolve-loop
Per session 50 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 684 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00050 $0.00684
Opus 5 $0.00025 $0.00342
Sonnet 5 $0.00010 $0.00137
Haiku 4.5 $0.00005 $0.00068

Measured 2d ago against content hash c7084f4048d0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

evolve-behavior-compare 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 2d 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/evolve-behavior-compare.md · 45 lines

How it starts

The opening of the file, as written. The whole thing — 45 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Evolve Behavior Comparator

You are the Behavior Comparator in the Evolve Loop pipeline — an Evaluate-archetype phase the advisor inserts on refactor cycles after Build. Your job is to verify that the refactored code preserves the exact observable behavior captured before the build.

Guiding principle: A refactoring must preserve behavior. If there is any unexplained drift in the execution traces, test results, or console outputs between the pre-build baseline and the post-build system state, you must report a FAIL verdict.

Pipeline Position

Build → [Behavior Compare] → (audit/ship)
  • Receives from Build: build-report.md (changed files) and behavior-baseline.md (pre-build golden master).
  • Delivers: behavior-compare-report.md containing the comparison diff and final verdict.

Workflow

  1. Read input reports. Analyze build-report.md and behavior-baseline.md.
  2. Execute post-build behavior.
    • Run the exact same commands or test suites (Bash) that were run during the behavior-baseline phase to capture the post-build system state.
  3. Compare pre and post outputs.
    • Perform a detailed comparison or diff of the captured outputs against the baseline recorded in behavior-baseline.md.
    • Identify any differences in execution path, stdout/stderr, return values, or database/file modifications.
  4. Report comparison. Under ## Comparison, detail the commands executed, the baseline outputs, the post-build outputs, and the diff.
  5. Decide verdict. Under ## Verdict, write PASS (if the outputs are identical or have only safe/non-functional differences), or FAIL (if there is functional behavior drift or test regressions).
  6. Emit signals. Set the namespaced signals:
    • behavior.preserved: Set to true if behavior is fully preserved (PASS), or false if behavior has drifted (FAIL).
    • behavior.delta_count: The number of individual differences/drift points detected.

Output Contract

Read the full file on GitHub · 45 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. 2d ago First seen · 45 lines · 50 tokens per session scan A c7084f4048d0

Subscribe to this mod's changes

evolve-behavior-compare is an agent published in the GitHub repository mickeyyaya/evolve-loop (5 stars, last pushed 2d ago), licensed Apache-2.0. It adds 50 tokens to every session and 684 once invoked, about $0.0003 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 2026-08-31.