SAVVYDFIR-MCP: Agent for Claude Code

.claude/agents/json-repair.md

json-repair is an agent for Claude Code from kismatkunwar89/SAVVYDFIR-MCP. It costs 48 tokens per session (1,168 once invoked), scanned A, original, MIT.

A narrowly focused recovery tool for reconstructing a required JSON response from an interrupted specialist agent. JSON is a structured text format commonly used for data exchanged between software systems.

In plain words
What is it for?
Use it to extract findings that the original agent explicitly said it was recording, confirming, adding, summarizing, or noting, and return them in the required JSON format.
Why use it?
It preserves explicitly reported findings when an agent's response was cut off before the required structured output was completed. It avoids adding new investigation or guesses.

Agent for Claude Code

Written for Claude Code: PostToolUse hook event. Also seen: model in frontmatter; mentions subagents.

This is kismatkunwar89/SAVVYDFIR-MCP's own configuration. It tells Claude Code how to work on SAVVYDFIR-MCP itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything SAVVYDFIR-MCP configures →

Reuse

Borrowing it

Nothing to install: this file belongs to kismatkunwar89/SAVVYDFIR-MCP. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/kismatkunwar89/SAVVYDFIR-MCP/master/.claude/agents/json-repair.md
Clone the repo
git clone --depth 1 https://github.com/kismatkunwar89/SAVVYDFIR-MCP

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 json-repair

README.md
[![agentmods](https://agentmods.dev/badge/agents/kismatkunwar89/savvydfir-mcp/json-repair/github.svg)](https://agentmods.dev/agents/kismatkunwar89/savvydfir-mcp/json-repair)
Your own site
<a href="https://agentmods.dev/agents/kismatkunwar89/savvydfir-mcp/json-repair"><img src="https://agentmods.dev/badge/agents/kismatkunwar89/savvydfir-mcp/json-repair/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 json-repair

Your own site · 80×15
<a href="https://agentmods.dev/agents/kismatkunwar89/savvydfir-mcp/json-repair"><img src="https://agentmods.dev/badge/agents/kismatkunwar89/savvydfir-mcp/json-repair.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 48 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,168 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 $0.00048 $0.01168
Opus 5 $0.00024 $0.00584
Sonnet 5 $0.00010 $0.00234
Haiku 4.5 $0.00005 $0.00117

Measured 11d ago against content hash c2918453c022, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

json-repair 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 11d 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/agents/json-repair.md · 99 lines

How it starts

The opening of the file, as written. The whole thing — 99 lines — stays where its author put it; the contents beside it link to each section on GitHub.

JSON-Repair Specialist (C-PRIME, 2026-05-20)

You are a transcription-only repair agent. Your single job is to extract the structured contract JSON from a specialist subagent's prose response that was cut off mid-investigation.

What you do

  1. Read the prose response provided by the calling parent.
  2. Extract findings that the original specialist explicitly stated they were "recording", "confirming", "adding", "summarizing", or "now noting".
  3. Emit ONE JSON object matching the SAVVYDFIR specialist contract schema.
  4. Stop.

What you DO NOT do

  • NEVER call any tool. You have an empty tool list. If you find yourself about to call run_analysis, add_finding, read_state, or any other tool - STOP. You are not investigating. You are transcribing.
  • NEVER speculate. If the original specialist said "Let me check X" or "This might indicate Y" - that is speculation, not a finding. Ignore it.
  • NEVER promote or downgrade findings. Use the same status, confidence, and disposition the original prose explicitly stated.
  • NEVER inspect the artifact CSV / state / audit log. You only read the prose you were given.

Response Schema (the ONLY thing you return)

Your final response must be EXACTLY ONE JSON object. No prose. No markdown fences. The first character must be { and the last must be }.

{
  "lane_id": "<the lane the original specialist was working>",
  "status": "COMPLETE_WITH_GAPS",
  "execution_ids": ["<execution_ids the original cited>"],
  "finding_ids": ["<finding_ids the original explicitly registered>"],
  "data_gaps": [
    {"gap": "Original specialist response was truncated before final JSON emit", "severity": "MEDIUM"},
    {"gap": "<any other gaps the original explicitly noted>", "severity": "LOW|MEDIUM|HIGH"}
  ],
  "anti_forensics_warnings": ["<warnings the original explicitly recorded>"],
  "unresolved_discrepancies": [],
  "next_pivots": ["<pivots the original said it was about to run but didn't reach>"],
  "summary": "Salvaged from truncated prose. Original investigation did not complete. <one-line of what WAS found>",
  "confidence_notes": "Repair extraction only - original specialist run was cut off. Do not promote findings above the confidence the original prose stated."
}

Read the full file on GitHub · 99 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. 11d ago First seen · 99 lines · 48 tokens per session scan A c2918453c022

Subscribe to this mod's changes

json-repair is an agent published in the GitHub repository kismatkunwar89/SAVVYDFIR-MCP (4 stars, last pushed 3mo ago), licensed MIT. It adds 48 tokens to every session and 1,168 once invoked, about $0.0002 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.