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.
git clone --depth 1 https://github.com/gonzalezpazmonica/saviaWrote 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/gonzalezpazmonica/savia/coherence-court)<a href="https://agentmods.dev/commands/gonzalezpazmonica/savia/coherence-court"><img src="https://agentmods.dev/badge/commands/gonzalezpazmonica/savia/coherence-court/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.
<a href="https://agentmods.dev/commands/gonzalezpazmonica/savia/coherence-court"><img src="https://agentmods.dev/badge/commands/gonzalezpazmonica/savia/coherence-court.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00025 | $0.00576 |
| Opus 5 | $0.00013 | $0.00288 |
| Sonnet 5 | $0.00005 | $0.00115 |
| Haiku 4.5 | $0.00003 | $0.00058 |
Grade A, and why
Coherence Court 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- Coherence Court — 100% identical, 0 lines differ
How it starts
The opening of the file, as written. The whole thing — 51 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/coherence-court — Transversal Consistency Audit
Runs the Coherence Court on a stage output against the premises registered for
the flow. 4 transversal judges (factual, scope, objectives, premise-drift) review
in parallel, producing a .coherence.crc verdict with a 0-100 score and a human
gate when the score falls under threshold.
Usage
/coherence-court --flow <flow> --stage-output <file>
/coherence-court --flow <flow> --stage-output <file> --register-only # solo registrar premisas
Flow
- Gate:
bash scripts/coherence-court.sh check --flow F --stage-output FILE— FAIL if flow has no premises (single-stage flow, nothing to compare). - Premises: ensure prior-stage premises are registered:
bash scripts/coherence-court.sh premises F add <kind> <content>(kind: fact|constraint|objective|decision). - Skeleton:
bash scripts/coherence-court.sh skeleton F FILE→.coherence.crc. - Judges: launch 4 subagents in parallel via
coherence-court-orchestrator:coherence-factual-judge: fact contradictions with prior stagescoherence-scope-judge: scope/constraint violationscoherence-objectives-judge: declared-objective contradictionscoherence-premise-drift-judge: silent premise drift
- Consolidate: score = 100 - (C×25 + H×10 + M×3 + L×1). Write
.coherence.crc. - Gate:
bash scripts/coherence-court.sh gate <score>→ PASS/CONDITIONAL/FAIL. - Report: discrepancies summary to the user.
Verdicts
| Score | Verdict | Next step |
|---|---|---|
| 90-100 | pass | Continuar flujo, revisión humana ligera |
| 70-89 | conditional | Revisar discrepancias antes de continuar |
| < 70 | fail | Puerta humana: NO continuar el flujo — humano decide |
Principles
- "Se delega la ejecución, nunca el criterio" — Coherence Court señala, no corrige.
- "Humano decide" — gate obligatorio cuando el score cae bajo umbral.
- CRIT-001 — premisas en texto plano local (
data/), sin salir de la máquina.
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 · 51 lines · 25 tokens per session scan A c47290a6e60b
Coherence Court is a command published in the GitHub repository gonzalezpazmonica/savia (50 stars, last pushed 2d ago), licensed MIT. It adds 25 tokens to every session and 576 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 2026-09-06.
Other commands, from other repositories
test
Design or run focused test validation for a task, bug, or diff.
verify
Verify a repository in fast or full mode with runner-first execution, deterministic stage order, and explicit PASS/FAIL/SKIP output.
simplify
Wrap-up refactoring — simplify code, eliminate duplication, preserve behavior.
tickets-implement
This is a user-invocable command. Execute it only because the current invocation explicitly selected it. It takes no arguments: every invocation processes the complete current project queue that is ready for implementation.
precommit-fast
Deprecated forwarding alias for fast pre-commit verification.
codex-test-gen
Deprecated alias for dhpk TDD test-generation mode.