Old Coder

Summary

Old Coder is an open-source skill that makes coding agents prove their work by requiring them to pass code through a structured gauntlet of checks, providing the user with a test plan and an evidence report instead of raw code. It requires no payment, and it runs as a skill added to various coding agents by dropping the skill folder into the appropriate agent skills directory. This tool is for application developers working with coding agents. Its documentation positions it alongside other agent skills for workflow and evidence management. The project was last updated via manual installation instructions, indicating active maintenance.

An old coder's strategy for the agent era: don't read the code — make it run the gauntlet. Evidence-first development skill for coding agents, inspired by Uncle Bob.

Command line

Values are placeholders taken from each option’s declared default. Nothing is executed here — the output shown is a recording of a run that already happened.

Release history

1. changes since 2026-07-27 Aug 18, 2026 · issue 008

   Adds Java, Scala, SQL, and Emacs Lisp support; new source-state binding, fail-closed gauntlet orchestration, and PYTHONDONTWRITEBYTECODE mutation isolation.

   └──▷ GET THIS VERSION
   copy

   $ git clone --branch commits-2026-07-27 https://github.com/AmazingAng/old-coder.git
   # already have the repo? check out this version:
   $ git checkout commits-2026-07-27

   └──▷ TRY IT
   Run mutation tests with bytecode caching disabled to prevent silent reuse of stale .pyc files between mutants of identical size written in the same second.
   copy

   $ PYTHONDONTWRITEBYTECODE=1 python demo-rate-limiter/tools/mutants.py

   Capture a reproducible, fail-closed source-state binding before running the gauntlet, so evidence can be tied to a specific, provenance-verified tree.
   copy

   $ python demo-rate-limiter/tools/source_state.py

   Verify that the gauntlet fails closed when a layer is omitted or a layer command breaks, using the negative-control harness.
   copy

   $ sh demo-rate-limiter/tools/test_gauntlet_orchestration.sh

   • ›Adds PYTHONDONTWRITEBYTECODE=1 via MUTANT_ENV in demo-rate-limiter/tools/mutants.py to prevent stale .pyc bytecode from silently reusing a prior mutant's compiled code between mutation runs.
   • ›Introduces demo-rate-limiter/tools/source_state.py, a new script that produces a deterministic, fail-closed binding for the demo source tree, withholding provenance when Git history is shallow.
   • ›Adds demo-rate-limiter/tools/test_gauntlet_orchestration.sh, a negative-control harness that verifies the gauntlet fails closed when a layer is omitted and that a broken layer command halts execution and names the failed layer.
   • ›Adds supply-chain and suite-health gauntlet layers to the orchestration set.
   • ›Extends language support to Java, Scala, SQL, and Emacs Lisp (PR #2).
   +4 moreshow less

   • ›Specifies fail-closed gauntlet orchestration: a skipped or missing layer causes the final audit to fail and name the offending layer.
   • ›Adds a VERIFY protocol step and two gauntlet rules derived from the independent verification experiment; verification is now explicitly source-state-specific and human-graded.
   • ›Splits gauntlet templates out of gauntlet.md and divides 'skipped layers' into three distinct categories.
   • ›Adds CONTRIBUTING.md defining the contribution bar and the rationale for the unusual review process here.