Ken Ranosa
Book a 15-min call

Agentic TDD Harness

Sole engineer. Designed, built, and evaluated the pipeline.

Timeframe

[Pending]

Timeframe: build date, run length, first-commit date.

Stack

Claude Code · parallel CLI agents · always-on hooks

Predates the playbook. The lessons here became it.

Loop engineering, before the name

[Pending]

Section held for the harness first-commit date.

The claim ships with a date or not at all.

Problem.

Agents write plausible code faster than one human can verify it.

Plausible is not verified.

Speed without verification is just risk arriving later.

I wanted ticket in, verified diff out. No babysitting.

[Pending]

Ticket source: the codebase the 101 tickets ran against.

Constraints.

One engineer. No second pair of eyes on demand.

Agents do not judge their own work. Every gate is external.

PRs cap at 100 authored LOC. Review speed is the budget.

No code beyond test scope. The tests define the diff.

Approach.

  1. 1

    Plan in parallel. Analyzer and test-scout waves feed 5 verifiers.

  2. 2

    Compete, then commit. N plans race; the winner absorbs expert fixes.

  3. 3

    Build on a leash. RED to GREEN to VERIFY; 2 attempts, then stop.

  4. 4

    Review in force. 9 parallel reviewers, then 14 pre-ship agents.

  5. 5

    Ship at zero findings. Auto-fix loops back to BUILD, 20 times max.

Principles: Context is King. Never Trust, Enforce. Validate Always.

Artifact.

The pipeline, distilled
ALWAYS-ON GUARDRAILS · HOOKS · NO-STUB · WAVE GATES · SPEC ALLOWLISTSPLANparallel CLI sessionsanalyzer + test-scout waves5-verifier consensus: 4/5, three roundsSELECTN plans competewinner absorbs expert correctionslosers discardedBUILD/tdd: RED → stub-immunity → GREEN→ VERIFY → SIMPLIFY → COMMIT2-attempt cap; no code beyond test scopeREVIEW9 parallel reviewersPRE-SHIP14 agents; expert criticsgate: 0 findings at every severityE2ESHIPPOST-SHIP/clear-pr-findings until APPROVEDadaptive learning from human reviewsAUTO-FIX→ BUILD20× MAX

The pipeline

PLAN       parallel CLI sessions
           analyzer + test-scout waves
           5-verifier consensus: 4/5, three rounds
SELECT     N plans compete
           expert corrections synthesized
           losers discarded
BUILD      /tdd: RED → stub-immunity → GREEN
           → VERIFY → SIMPLIFY → COMMIT
           2-attempt cap; no code beyond test scope
REVIEW     9 parallel reviewers
PRE-SHIP   14 agents; expert critics; auto-fix
           gate: 0 findings at every severity
           auto-fix loops back to BUILD, 20x max
E2E → SHIP
POST-SHIP  /clear-pr-findings until APPROVED
           adaptive learning from human reviews

Always-on guardrails: pre-write, pre-bash, and post-write hooks;

no-stub enforcement; wave gates; spec allowlists.

93/101

tickets completed to a verified PR

0

findings allowed at ship, at every severity

20x

hard cap on auto-fix loop-backs

100

authored-LOC cap per PR

[Pending]

Example diff and PR artifact.

Results.

93/101

Tickets in, verified PRs out. One pipeline, end to end.

8

Tickets failed. The write-up is owed, not spun.

[Pending]

Definition of verified PR: gates and review criteria.

Failure analysis: why 8 tickets missed.

Learnings.

The lessons became the playbook:

small slices, external gates, hard stops.

Verification is the product. The code is a byproduct.

[Pending]

What I would change: gated on the failure analysis of the 8 misses.

How can I help?

Ken Ranosa

Senior Full-Stack Engineer

Test Infrastructure Edge

Remote from Makati, Philippines · GMT+8 · Available now

Need a senior engineer who can ship product and improve release confidence?

ken.ranosa@gmail.com

linkedin.com/in/mranosa

Remote from Makati, Philippines · GMT+8