session-audit

A command that records how a coding session behaved, including tool calls, edits, errors, retries, and reasoning, then sends that record to Perplexity for an independent review. Perplexity is an external research and answer service.

In plain words
What is it for?
Use it to review a difficult or unproductive Claude Code session and find patterns in its mistakes, repeated attempts, or inefficient actions.
Why use it?
It helps identify when an AI coding session is going in circles or working inefficiently. The result is a retrospective scorecard and list of findings, not an implementation plan.

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/intellegix/intellegix-code-agent-toolkit/session-audit
Clone the repo
git clone --depth 1 https://github.com/intellegix/intellegix-code-agent-toolkit
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 4,425 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 $0.00000 $0.04425
Opus 5 $0.00000 $0.02212
Sonnet 5 $0.00000 $0.00885
Haiku 4.5 $0.00000 $0.00443

Measured 2d ago against content hash 5a56bf0987db, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

session-audit 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/session-audit.md · 345 lines

How it starts

The opening of the file, as written. The whole thing — 345 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/session-audit — Self-Diagnostic Performance Audit via Perplexity

Capture session behavior (tool calls, edits, errors, retries, reasoning) and send to Perplexity for independent performance review. Use when Claude seems to be struggling, going in circles, or operating suboptimally.

Cost: Free — uses Perplexity login session only (Playwright browser automation).

This is a RETROSPECTIVE AUDIT, not a planner. It outputs a scorecard and findings directly. Do NOT enter plan mode. Do NOT ask the user questions before completing Step 0. Compile evidence silently and execute.

MANDATORY CONTEXT PREAMBLE — EVERY QUERY, NO EXCEPTIONS: Every single query sent to Perplexity MUST begin with the following preamble block. This is a hard rule — never omit it, never paraphrase it, never move it to a footnote. It goes at the TOP of every query, before any other content:

[ENVIRONMENT CONTEXT — READ FIRST]
This project is being developed using Claude Code, Anthropic's official CLI tool for Claude (claude.ai/claude-code). The developer uses a Claude Max subscription and works entirely in the terminal via the `claude` CLI command. Claude Code is an agentic coding assistant that reads/writes files, runs terminal commands, searches codebases, and executes multi-step development tasks autonomously. All code generation, refactoring, debugging, and project management happens through Claude Code's conversation interface — there is no IDE or GUI involved. Responses should account for this workflow: recommend CLI-compatible tools, terminal-based solutions, and approaches that work well with an AI coding agent operating in a command-line environment.
[END ENVIRONMENT CONTEXT]

Input

$ARGUMENTS = Optional flags. Supports:

  • Empty (default): Audit from last checkpoint or last 8 hours
  • --full: Bypass checkpoint, audit all available session history (keeps diff/log caps)

Workflow

Step 0: Compile Session Evidence — MANDATORY, SILENT

Before doing ANYTHING else, compile evidence of what happened this session. Do NOT ask the user any questions during this step — proceed silently and autonomously.

Read the full file on GitHub · 345 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 · 345 lines · 0 tokens per session scan A 5a56bf0987db

Subscribe to this mod's changes

session-audit is a command published in the GitHub repository intellegix/intellegix-code-agent-toolkit (57 stars, last pushed 8d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 4,425 tokens. 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 2026-08-30.