Summary

knot-rider was scaffolded from rsr-template-repo but never instantiated — {{PLACEHOLDER}} tokens survived throughout, including in the .zig sources, so the Zig FFI seam genuinely did not parse. This PR cures the instantiation for real.

• Identity placeholders substituted across ~80 files ({{PROJECT_NAME}}, {{OWNER}}, {{REPO}}, {{AUTHOR}}, {{FORGE}}, {{LICENSE}}, etc.) — real values used throughout, intentional generic scaffolding (docs-template/, ADR templates, K9 example configs, native Justfile {{recipe}} interpolation) left untouched.
• .github/settings.yml landmine fixed: an unsubstituted name: "{{REPO}}" under the probot/settings config would have made the app try to rename the repo to the literal string {{REPO}} on the next push to main.
• Zig FFI seam now genuinely builds and tests: {{project}}_* → knot_rider_*, plus two real Zig 0.16 incompatibilities fixed along the way (opaque types can no longer carry fields; callconv(.C) → callconv(.c)). build.zig was previously inert scaffolding with zero real build targets — rewrote it to actually compile main.zig as a static library and wire zig build test to the real test suites. zig fmt --check, zig build, and zig build test all pass (4/4 tests).
• .machine_readable/6a2/{CLADE,ECOSYSTEM,STATE}.a2ml still described rsr-template-repo's own identity/purpose/"95% complete, production maturity" claim — hardcoded wrong content, not a {{}} token. Fixed the fields that feed the CLAUDE.md arrival-pack generator, then regenerated CLAUDE.md for real via just claude-md rather than hand-editing the generated region.
• Justfile TODO/echo stubs rewired to real commands: build/test/fmt/fmt-check → real zig build/zig build test/zig fmt; lint → idris2 --typecheck abi.ipkg; e2e/aspect → bash tests/e2e.sh / bash tests/aspect_tests.sh; deps → checks the Zig toolchain. Deleted run/run-verbose/bench/readiness recipes rather than leave echo "..." fake-pass stubs — this repo is a C-ABI library with no application entry point and no bench/readiness suite exists.
• Two real false positives fixed in tests/aspect_tests.sh that the newly-wired aspect recipe surfaced: documentation prose describing banned patterns (sorry, Admitted) was being flagged as a violation, and Zig build-cache files were being scanned for SPDX headers. just test-all / just quality are genuinely green now, not just newly-wired-but-red.
• Removed obsolete template machinery: scripts/validate-template.sh, tests/e2e/template_instantiation_test.sh, benches/template_bench.sh, .machine_readable/ai/PLACEHOLDERS.adoc. No workflow referenced any of them (re-verified).
• README/EXPLAINME/AFFIRMATION rewritten honestly.

This project's purpose remains undocumented

This is deliberate, not an oversight. knot-rider's specific technical purpose is not documented anywhere in this repository or the wider hyperpolymath estate registry (checked: reposystem tool registry, verisimdb-data, gitbot-fleet, estate memory). The only repo-specific content that exists is GitHub's own description, a knowing pun on the Knight Rider intro monologue:

A shadowy flight into the dangerous world of a man, who is either permanently or very often undone by circumstance.

That's wordplay, not a spec. README.adoc says this plainly rather than inventing a product story.

Known, out-of-scope findings (not fixed here)

• The literal string rsr-template-repo (not a {{}} placeholder, so outside this PR's defined scope) still appears as leftover self-referential text in ~40 files under .machine_readable/contractiles/, docs/onboarding/, docs/status/, .machine_readable/coaptation/, etc. I fixed the highest-impact instances (the ones feeding CLAUDE.md, the Justfile's user-facing doctor/tour output, and de-duplicated .machine_readable/contractiles/Justfile back into a hardlink of the root Justfile per its own dust-check). A full sweep of this second bug class would roughly double the size of this change and is a reasonable follow-up.
• docs/status/TEST-NEEDS.adoc describes rsr-template-repo's own historical test-suite development (not knot-rider's) — left as-is, flagged as stale.
• idris2 --typecheck abi.ipkg was verified to pass in this environment (exit 0, warnings only), but this wasn't part of the original ask and toolchain availability may differ elsewhere.

Test plan

• zig fmt --check src/interface/ffi — exit 0
• cd src/interface/ffi && zig build — exit 0
• cd src/interface/ffi && zig build test — exit 0, 4/4 tests pass
• just quality (fmt-check + lint + test) — exit 0
• just test-all (test + e2e + aspect) — exit 0
• just claude-md && just validate-claude-md — in sync
• Edited workflow YAML re-parsed with python3 -c "import yaml; yaml.safe_load(...)"
• Grepped for lingering .github/workflows/*.yml references to the deleted template-machinery scripts — none found

🤖 Generated with Claude Code