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/xsovad06/sova/specgit clone --depth 1 https://github.com/xsovad06/sovaWhat 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 | $0.00014 | $0.02082 |
| Opus 5 | $0.00007 | $0.01041 |
| Sonnet 5 | $0.00003 | $0.00416 |
| Haiku 4.5 | $0.00001 | $0.00208 |
Grade A, and why
spec 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 yesterday.
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 — 275 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Specification
Produce a structured specification document for a task before development starts. Shifts architectural and UX decisions to the cheap planning phase instead of the expensive coding phase.
# Benchmark logging (entry)
bash .claude/benchmark/log.sh "spec_start" "" "" 2>/dev/null || true
Task: $ARGUMENTS
Instructions
Step 1: Fetch the Task
If $ARGUMENTS is a text description (not a number or ticket key), use it directly as the problem statement and skip to Step 2.
Determine the task source by reading sova.toml (if it exists) and checking [task_source] type.
GitHub (default, or no sova.toml):
gh issue view $ARGUMENTS --json number,title,body,labels,milestone
JIRA (task_source.type = "jira"):
jira issue view $ARGUMENTS --plain
Extract: title, description, status, linked/blocked tickets, components.
Save the original description verbatim -- it will be preserved in the spec.
Early exit -- already implemented: if the issue body contains a Research section that concludes the issue is already fully implemented (look for phrases like "already implemented", "already complete", "no remaining work"), write a minimal spec file to .claude/specs/{issue-number}-already-implemented.md with the following content, then stop -- do not explore the codebase or write a full spec:
# Spec: {Issue title}
**Issue**: #{number}
**Status**: approved
**Created**: {YYYY-MM-DD}
**Complexity**: trivial
## Problem
Already implemented per research findings.
## Solution
No changes needed. The research step confirmed this issue is already fully implemented.
Step 2: Read Project Context
Read whatever exists -- skip files that are missing:
CLAUDE.mdandAGENTS.mdfor project conventions.claude/rules/architecture.mdfor component overview and design decisions.claude/agent-memory/cookbook.mdfor known patterns and past mistakes
Step 3: Explore the Codebase
Based on the issue, identify the affected areas:
- Find the modules, files, and functions that will need changes
- Read existing patterns in those areas -- match them, don't invent new ones
- Identify reusable components, utilities, and test fixtures
- Check for related prior work (similar features, relevant tests)
- Find the closest precedent -- which prior task or file implements something similar? Note specific file paths and line numbers as the pattern reference.
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.
- yesterday First seen · 275 lines · 14 tokens per session scan A dc6528bf8486
spec is a command published in the GitHub repository xsovad06/sova (2 stars, last pushed 2d ago), licensed Apache-2.0. It adds 14 tokens to every session and 2,082 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
align
Unified alignment command (--project, --docs, --retrofit, --content).
dashboard
Phase 3 monitoring — System health, execution insights, and velocity analytics.
doctor
Check the health of a KARIMO installation, identify issues, and provide actionable recommendations.
greptile-review
Execute the full Greptile review cycle on a PR, looping until score meets threshold or circuit breaker triggers.
update
Check for and apply KARIMO updates from GitHub releases.
implement-fix
Minimal pipeline for test-fixing tasks.