against

A command that compares two saved scorecards for the same skill and shows how each measured area changed.

In plain words
What is it for?
Comparing a baseline review with a later review, displaying per-dimension score differences, and using the overall result as a continuous-integration gate.
Why use it?
It makes it easier to see whether a skill improved, regressed, or stayed unchanged. It can also return a failure status when the overall score gets worse, allowing CI to block a change.

Command

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.

agentmods
npx agentmods add commands/sattyamjjain/proofloop/against
Clone the repo
git clone --depth 1 https://github.com/sattyamjjain/proofloop
Per session 14 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 695 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00014 $0.00695
Opus 5 $0.00007 $0.00347
Sonnet 5 $0.00003 $0.00139
Haiku 4.5 $0.00001 $0.00069

Measured 2d ago against content hash aa751a9fc4b0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

against 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 2d 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.

commands/against.md · 74 lines

How it starts

The opening of the file, as written. The whole thing — 74 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/against — Diff-Aware Score Delta

Renders a Unicode table comparing two scorecards for the same skill — per-dimension deltas plus a composite verdict (IMPROVED / REGRESSED / FLAT). Exits non-zero on composite regression so CI can gate on it.

Arguments

  • skill-name (required): Which skill's history to compare.
  • --baseline-index N (optional, default -2): Index into the skill's scorecards sorted by timestamp ascending. Negative indices count from the end, so -2 is the penultimate run (one before latest).
  • --target-index N (optional, default -1): Same indexing as the baseline. -1 is the latest run.

What to Do

  1. Invoke the CLI:

    python3 skills/judge/scripts/against.py \
      --skill {skill-name} \
      --scores-dir skills/judge/scores \
      [--baseline-index N] [--target-index N]
    
  2. The script reads every {skill-name}_*.json file from the scores directory, sorts by timestamp, selects the two indices, and prints a delta table like:

    ╭────────────────────────────╮
    │ Proofloop delta: code-review │
    ╰────────────────────────────╯
    baseline: 2026-04-10T12:00:00Z  composite 7.80/B
    target:   2026-04-18T12:00:00Z  composite 8.60/A-
    
    dimension        before    after    delta
    ─────────────────────────────────────────────
    correctness          8        9    +1.0 ↑
    completeness         7        8    +1.0 ↑
    adherence            8        9    +1.0 ↑
    actionability        8        8    +0.0 →
    efficiency           7        8    +1.0 ↑
    safety              10       10    +0.0 →
    consistency          5        8    +3.0 ↑
    ─────────────────────────────────────────────
    composite         7.80     8.60   +0.80 ↑
    
    Proofloop: IMPROVED (+0.80)
    
  3. Exit semantics:

    • Exit 0 — improved or flat (≥ -0.05 composite delta).
    • Exit 1 — fewer than 2 scorecards, or indices out of range.
    • Exit 2 — regression (composite dropped > 0.05). CI gate.

Examples

/against code-review                           # penultimate vs. latest
/against code-review --baseline-index 0        # first-ever vs. latest
/against code-review --baseline-index -5 --target-index -1  # last 5 runs

Read the full file on GitHub · 74 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. 2d ago First seen · 74 lines · 14 tokens per session scan A aa751a9fc4b0

Subscribe to this mod's changes

against is a command published in the GitHub repository sattyamjjain/proofloop (5 stars, last pushed 2mo ago), licensed MIT. It adds 14 tokens to every session and 695 once invoked, about $0.0001 per session on Opus 5. 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 2026-08-31.