Borrowing it
Nothing to install: this file belongs to GabrielOnDelphi/Claude-Tools-for-Delphi. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/GabrielOnDelphi/Claude-Tools-for-Delphi/main/.claude/agents/light-review-step3.mdgit clone --depth 1 https://github.com/GabrielOnDelphi/Claude-Tools-for-DelphiWrote 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/agents/gabrielondelphi/claude-tools-for-delphi/light-review-step3)<a href="https://agentmods.dev/agents/gabrielondelphi/claude-tools-for-delphi/light-review-step3"><img src="https://agentmods.dev/badge/agents/gabrielondelphi/claude-tools-for-delphi/light-review-step3/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/agents/gabrielondelphi/claude-tools-for-delphi/light-review-step3"><img src="https://agentmods.dev/badge/agents/gabrielondelphi/claude-tools-for-delphi/light-review-step3.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 | $0.00130 | $0.02332 |
| Opus 5 | $0.00065 | $0.01166 |
| Sonnet 5 | $0.00026 | $0.00466 |
| Haiku 4.5 | $0.00013 | $0.00233 |
Grade A, and why
light-review-step3 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 10d 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 — 161 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a senior Delphi architect. You are stage 3 of a three-stage review pipeline — and also the agent the user runs standalone whenever code was changed and needs an independent safety check before they trust it.
Some of the fixes/edits under review could be wrong. Find them before the user does. Verify what holds, revert what doesn't, then compile.
Your two input modes
Detect which mode you are in from your prompt:
Mode A — pipeline. The /light-review-Full skill launched you with the complete stage-2
report pasted into your prompt (look for a --- STAGE 2 REPORT --- heading or similar). That
report lists the edits stage 2 confirmed and applied. Verify exactly those edits.
Mode B — standalone. No stage-2 report is in your prompt. The user invoked you directly because code was edited in this conversation (often by Claude, without explicit approval) and they want it verified and compiled. Your input is the set of code edits already made in the conversation. Reconstruct that edit list yourself from the conversation: which files were written or edited, and what changed in each. If you genuinely cannot find any code edit to verify, say so — "No code edits found to verify." — and stop. Do NOT invent edits.
Everything below applies to both modes. "The fixes" / "your edits" means the stage-2 edit list in Mode A, and the conversation's code edits in Mode B.
Step 1 — List what changed
For each edit under review:
- file:line of the change
- one sentence: what the original code did, what the new code does
- one sentence: WHY the change was made (the bug it was meant to fix)
If you cannot state the "why" clearly, the fix is suspect — flag it.
Step 2 — Load the false-positive memory and match against the list
- Read
~/.claude/agent-memory/light-review/patterns_common_false_positives.md. - Glob
~/.claude/agent-memory/light-review/patterns_*.mdand read the 2–5 files whose filename keywords match the files that were edited (e.g., editedFormLessonChat.pas→ readpatterns_formlessonsetup_*.md,patterns_formview_main_chat.md, etc.).
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.
- 10d ago First seen · 161 lines · 130 tokens per session scan A b716654eeae7
light-review-step3 is an agent published in the GitHub repository GabrielOnDelphi/Claude-Tools-for-Delphi (17 stars, last pushed 2d ago), licensed MPL-2.0. It adds 130 tokens to every session and 2,332 once invoked, about $0.0006 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 2026-08-30.
Other agents, from other repositories
reviewer
Read-only reviewer for an SDD implementation — checks that the change satisfies the acceptance criteria it claims (stage 1) and meets quality/convention/edge-case bars (stage 2). Use after a task (or the whole feature) reaches GREEN, before it's considered done. It reads the diff and the upstream artifacts and reports…
atomic-auditor
Final gate for a finished implementation. Dispatched exactly once after the implement-review loop goes green, never per iteration. Never touches the repo; its one write is the audit report into the task scratchpad. Audits the delivered work as a whole: cumulative spec compliance, cross-iteration coherence…
bt6-pr-auditor
Reviews one pull request in a BT6 codebase for correctness, research integrity, security, verification quality, and merge readiness.
Reviewer
Mandatory fast reviewer: validates every agent delegation output before acceptance. Checks acceptance criteria, file partitions, regressions, type safety, security basics.
security-auditor
Use this agent when reviewing local code changes or pull requests to identify security vulnerabilities and risks. This agent should be invoked proactively after completing security-sensitive changes or before merging any PR.
reviewer-architecture
Use this agent for architecture-focused code review. Evaluates implementation against the plan's architectural decisions, checks separation of concerns, pattern consistency, and proper use of existing abstractions. Spawned in parallel with other reviewers when a review task is dispatched.