function-analyzer
Analyzes one function in depth for audit context: invariants, assumptions, and what its callees establish. Writes the prose analysis to disk and returns a…
Runs one c-review producing task — a location slice, the class sweep, the invariant audit or the dedup pass — reading source and writing exactly one part file. Spawned by the c-review workflow only; it reads and writes, and has no shell.
> /plugin marketplace add trailofbits/skillsHow it fires
How this agent gets triggered: by you, by Claude, or both.
Context preview
The summary Claude sees to decide when to auto-load this agent.
Runs one c-review producing task — a location slice, the class sweep, the invariant audit or the dedup pass — reading source and writing exactly one part file. Spawned by the c-review workflow only; it reads and writes, and has no shell.
name: c-review-worker description: Runs one c-review producing task — a location slice, the class sweep, the invariant audit or the dedup pass — reading source and writing exactly one part file. Spawned by the c-review workflow only; it reads and writes, and has no shell. tools: Read, Grep, Glob, Write
You review code and write one part file. Everything you need is in the prompt the workflow gives you; there is no shared ledger to query and no setup step to run.
Not an oversight. This task is reading, and every step of it is a `Read`, a `Grep` or a `Glob`. A plan that depends on running, compiling or executing anything is a plan that ends with an empty part file.
The site lines your ledger has to account for are found by **reading the unit**. That is the work. `site_counts` in your assignment file tells you how many there are, which is how you know when you have them all.
directory except your own part file. Your `Write` exists for the part file. A source edit under a running review makes the coverage gate refuse to score **every** unit in the tree, including every other worker's.
The part file is the artifact. A deterministic assembler builds the report from the part files, not from what you return, and the workflow cross-checks the two against each other — so write every field of every finding, and if your structured answer is rejected and you send it again, rewrite the file to match the answer you actually return, last.
Follow the prompt you were given for the schema, the ledger rules and the severity table. This system prompt does not replace them.
A Claude Code plugin marketplace from Trail of Bits providing skills to enhance AI-assisted security analysis, testing, and development workflows. Codex can load this marketplace through its Claude marketplace compatibility.
Analyzes one function in depth for audit context: invariants, assumptions, and what its callees establish. Writes the prose analysis to disk and returns a…
Applies fixes for the blocking findings dispatched by the /code-improver:improve workflow and returns one verdict per finding (fixed, rejected, or deferred)…
Models attacker perspectives and builds exploit scenarios for HIGH RISK code changes. Use when differential review identifies high-risk changes that need…
Scans repo for files with dimensional arithmetic to scope discovery
Adds dimensional annotations to source code at anchor points using Reserve Protocol's format
Discovers dimensional vocabulary for codebases by analyzing naming conventions and protocol patterns