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.
npx agentmods add commands/sattyamjjain/proofloop/judge-configgit clone --depth 1 https://github.com/sattyamjjain/proofloopWrote 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.
[](https://agentmods.dev/commands/sattyamjjain/proofloop/judge-config)<a href="https://agentmods.dev/commands/sattyamjjain/proofloop/judge-config"><img src="https://agentmods.dev/badge/commands/sattyamjjain/proofloop/judge-config.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00012 | $0.00806 |
| Opus 5 | $0.00006 | $0.00403 |
| Sonnet 5 | $0.00002 | $0.00161 |
| Haiku 4.5 | $0.00001 | $0.00081 |
Grade A, and why
judge-config 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 5d 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.
How it starts
The opening of the file, as written. The whole thing — 105 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/judge-config — Manage Auto-Judge Settings
View and modify the Proofloop configuration.
Subcommands
/judge-config— View current configuration (pretty-printed)/judge-config add-always SKILL— Add a skill to the auto-judge "always" list/judge-config add-never SKILL— Add a skill to the auto-judge "never" list/judge-config remove SKILL— Remove a skill from both always and never lists/judge-config threshold N— Set the blocking threshold (0.0-10.0)/judge-config enable— Enable auto-judging/judge-config disable— Disable auto-judging
What to Do
- Read
judge-config.jsonfrom the project root - If no subcommand: display the full config in a readable format:
Proofloop Configuration
========================
Auto-Judge: ENABLED
Threshold: 5.0 (skills scoring below this are BLOCKED)
Always Auto-Judge:
- code-review
- security-scan
- feature-dev
- debugging
- codebase-analyzer
- webapp-testing
Never Auto-Judge:
- format
- commit
- fix-imports
- undo
- fix-todos
- remove-comments
- docs
- session-start
- session-end
Manual-Only (not in either list):
→ Everything else requires /judge for evaluation
Scoring Weights (global — per-rubric overrides apply when a
`<rubric>.weights.json` sidecar exists):
Correctness: 25%
Completeness: 20%
Adherence: 15%
Actionability: 15%
Efficiency: 10%
Safety: 10%
Consistency: 5%
Tokenizer Baselines (efficiency length-threshold multipliers):
default: 1.0
claude-opus-4-7: 1.35
claude-sonnet-4-6: 1.0
claude-haiku-4-5: 1.0
- If subcommand provided: modify judge-config.json accordingly
- Validate inputs (threshold must be 0-10, skill names must be valid)
- Remove from conflicting lists when adding (e.g., adding to "always" removes from "never")
- Confirm the change to the user
Weight validation
When the user edits scoring.dimensions directly, Proofloop enforces the
weight-sum-to-1.0 invariant at load time. Non-conforming configs are
rejected with a stderr warning and the scorer falls back to default
weights instead of silently producing inflated composites. Run
python3 skills/judge/scripts/score.py --config judge-config.json ...
once after an edit to confirm the warning doesn't fire.
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.
- 5d ago First seen · 105 lines · 12 tokens per session scan A b0364187952f
judge-config is a command published in the GitHub repository sattyamjjain/proofloop (5 stars, last pushed 2mo ago), licensed MIT. It adds 12 tokens to every session and 806 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.
Other commands, from other repositories
land
Cadence-tick autonomous PR babysitter (CI-fix, resolve, converge, merge, close, release).
sync
Manually trigger plan-sync to update downstream task specs after implementation drift.
uninstall
Remove flow-next files from project.
test
Run Postman collection tests, analyze results, diagnose failures, and suggest fixes.
routine
Execute one maintenance routine defined in scv/routines/ .md (task + guardrails + exit-criteria contract), or list defined routines. SCV never schedules — pair with host features like /loop or cron yourself. Use whenever the user asks to run a recurring maintenance task, or asks what routines exist — not only when…
t800-onboard
Для чата с новичками. Показывает, что настроено в Cursor (global + local), и что умеет отдел T-800.