Version Your Harness and Track Behavior

Treat prompts, tools, and loop logic as versioned, changelogged artifacts - so behavior changes are decisions, not surprises.

0 runs0 forks0

Chain

Inputs

Step 1

flow-version-your-harness-track-behavior-1-extract-behavior-into-versioned-artif

Step 2

flow-version-your-harness-track-behavior-2-start-the-behavior-changelog

Step 3

flow-version-your-harness-track-behavior-3-build-the-smoke-eval-and-behavioral-d

Step 4

flow-version-your-harness-track-behavior-4-prove-the-loop-with-a-real-change

Final output

No inputs required — hit Run to stream each step in order.

Each step streams from its model and feeds the next. BYOK — you pay your provider directly. Add a key in Settings if a step needs one.

Step 1 · You can only track what has an identity.

flow-version-your-harness-track-behavior-1-extract-behavior-into-versioned-artif

Claude Sonnet 4.6→ {{step_1}}

Awaiting upstream steps…

Step 2 · Write the diff a user would notice, not the diff git shows.

flow-version-your-harness-track-behavior-2-start-the-behavior-changelog

Claude Sonnet 4.6→ {{step_2}}

Awaiting upstream steps…

Step 3 · Ten fixed tasks make drift visible in minutes.

flow-version-your-harness-track-behavior-3-build-the-smoke-eval-and-behavioral-d

Claude Sonnet 4.6→ {{step_3}}

Awaiting upstream steps…

Step 4 · One tracked change end to end validates the whole system.

flow-version-your-harness-track-behavior-4-prove-the-loop-with-a-real-change

Claude Sonnet 4.6→ {{step_4}}

Awaiting upstream steps…

Treat prompts, tools, and loop logic as versioned, changelogged artifacts - so behavior changes are decisions, not surprises.

1
You can only track what has an identity.
Runs flow-version-your-harness-track-behavior-1-extract-behavior-into-versioned-artif on Claude Sonnet 4.6
2
Write the diff a user would notice, not the diff git shows.
Runs flow-version-your-harness-track-behavior-2-start-the-behavior-changelog on Claude Sonnet 4.6
3
Ten fixed tasks make drift visible in minutes.
Runs flow-version-your-harness-track-behavior-3-build-the-smoke-eval-and-behavioral-d on Claude Sonnet 4.6
4
One tracked change end to end validates the whole system.
Runs flow-version-your-harness-track-behavior-4-prove-the-loop-with-a-real-change on Claude Sonnet 4.6

•Recurring tasks you currently copy-paste between tools.
•When a single prompt isn't enough — you need a chain where output(n) feeds prompt(n+1).
•Fork and adapt if your inputs differ materially from the author's use case.

Claude Sonnet 4.6