champion-premise-false-evidence

champion-premise-false-evidence is a command for Claude Code from rjwalters/kicad-tools. It costs 0 tokens per session (1,177 once invoked), scanned A, original, MIT.

A reference for deciding whether the main factual claim in a proposal is false within the Champion issue workflow. It defines the evidence checks used when writing the finding and when deciding whether an issue may be closed.

In plain words
What is it for?
Use it to implement or review premise-false findings and the final close check in Champion's issue-promotion process.
Why use it?
It prevents false conclusions caused by reading symbolic links as if they were document contents. This matters when automated searches appear empty even though the referenced documentation is present.

Command for Claude Code

Written for Claude Code: installed under .claude/. Also seen: positional $N argument.

Good fit Use it to implement or review premise-false findings and the final close check in Champion's issue-promotion process.

Compare 6 commands from other repositories ↓
Install with agentmods
npx agentmods add commands/rjwalters/kicad-tools/champion-premise-false-evidence
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.

Clone the repo
git clone --depth 1 https://github.com/rjwalters/kicad-tools

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for champion-premise-false-evidence

README.md
[![agentmods](https://agentmods.dev/badge/commands/rjwalters/kicad-tools/champion-premise-false-evidence/github.svg)](https://agentmods.dev/commands/rjwalters/kicad-tools/champion-premise-false-evidence)
Your own site
<a href="https://agentmods.dev/commands/rjwalters/kicad-tools/champion-premise-false-evidence"><img src="https://agentmods.dev/badge/commands/rjwalters/kicad-tools/champion-premise-false-evidence/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for champion-premise-false-evidence

Your own site · 80×15
<a href="https://agentmods.dev/commands/rjwalters/kicad-tools/champion-premise-false-evidence"><img src="https://agentmods.dev/badge/commands/rjwalters/kicad-tools/champion-premise-false-evidence.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,177 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00000 $0.01177
Opus 5.5 $0.00000 $0.00471
Sonnet 5.5 $0.00000 $0.00235
Haiku 4.5 $0.00000 $0.00118

Measured 24d ago against content hash db11d22a210e, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

champion-premise-false-evidence 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 24d 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.

.claude/commands/loom/champion-premise-false-evidence.md · 81 lines

How it starts

The opening of the file, as written. The whole thing — 81 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Champion: premise-false evidence rules (#8593)

Reference for the two places champion-issue-promo.md decides whether a proposal's central factual claim is false: "The premise-false finding kind" (writing the finding) and "Premise-false close gate" (re-running it on the N=2 pass, the one place Champion may close an issue). The rules themselves are stated there; this file carries the resolver and the evidence table they point at.

What went wrong (#8494)

On 2026-09-21 the close gate closed a sound proposal not planned. Both of its recurring findings were "verified via" a grep of a .loom/docs/*.md path against origin/main, each returning no output — and both premises were true at that commit. Since #7842 every .loom/docs/*.md with a defaults/docs/ counterpart is a symlink: its tree entry is mode 120000, so reading it through git (git show origin/main:<path>, git cat-file -p, git archive) yields the link-target string (../../defaults/docs/x.md), never the document.

Every grep/wc over that read comes back empty whatever the document says. The gate's own escape hatch — "if any re-run check now shows the premise IS true, go to Step 1" — could not fire either, because the N=2 re-run was told to re-run the cited check verbatim, so it reproduced the identical false negative.

Resolve a cited path before checking it

resolve_main_path() {   # the blob git will actually read, bounded to 3 hops
  local p="$1" i
  for i in 1 2 3; do
    [ "$(git ls-tree origin/main -- "$p" | awk '{print $1}')" = 120000 ] || break
    p="$(dirname "$p")/$(git show "origin/main:$p")"   # relative to the LINK's dir
    # collapse "." / ".." lexically — the target is a path IN A COMMIT, so
    # realpath/readlink (which stat the working tree) are both wrong here
    p="$(printf '%s' "$p" | awk -F/ '{n=0;for(i=1;i<=NF;i++){if($i=="."||$i=="")continue;
      if($i==".."){if(n>0)n--;continue}a[++n]=$i}s=a[1];for(i=2;i<=n;i++)s=s"/"a[i];print s}')"
  done
  printf '%s\n' "$p"
}

P=$(resolve_main_path ".loom/docs/transcript-token-ingest.md")
# → defaults/docs/transcript-token-ingest.md
git show "origin/main:$P" | grep -c '30-day fuse'   # 3, not (no output)

Read the full file on GitHub · 81 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. 24d ago First seen · 81 lines · 0 tokens per session scan A db11d22a210e

Subscribe to this mod's changes

champion-premise-false-evidence is a command published in the GitHub repository rjwalters/kicad-tools (70 stars, last pushed today), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,177 tokens. 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 2026-09-23.

Related

Other commands, from other repositories