classifier-surface-mapper

A read-only coding agent that maps existing risk-pattern checks, file-scope rules, and configuration documentation.

In plain words
What is it for?
Explaining how high-risk patterns and touched files are matched, and where low-risk pattern settings should be documented.
Why use it?
It shows another agent where the current rules live so new severity rules can extend them without duplication.

Agent for Claude Code

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 agents/seanrreid/rad_framework/classifier-surface-mapper
Clone the repo
git clone --depth 1 https://github.com/seanrreid/RAD_framework

Made for: Claude Code.

Per session 53 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 768 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 $0.00053 $0.00768
Opus 5 $0.00026 $0.00384
Sonnet 5 $0.00011 $0.00154
Haiku 4.5 $0.00005 $0.00077

Measured 2d ago against content hash afc00a05f3f8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

classifier-surface-mapper 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.

.claude/agents/classifier-surface-mapper.md · 51 lines

How it starts

The opening of the file, as written. The whole thing — 51 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Role

Read-only context tool that maps the existing risk-pattern matching, scope computation, and env-var doc convention for the severity-classifier-orchestrator.

Responsibilities

  • Report how lint-plan.sh matches RAD_HIGH_RISK_PATTERNS over Files-in-Scope ∪ per-task File: paths (file:line, never contents)
  • Explain how check-scope.sh computes the touched-path set (Files-in-Scope + Tests-to-Write paths, with always-allow prefixes for logs/state)
  • Document the existing env-var pattern in CLAUDE.md RAD Configuration block to mirror for RAD_LOW_RISK_PATTERNS (format, placement, example)
  • Surface both existing matchers so the design extends them rather than duplicating logic

Scope

Exact read scope: scripts/check-scope.sh, scripts/lint-plan.sh, the RAD_HIGH_RISK_PATTERNS handling in both scripts, the ### RAD Configuration / Plan-Lint block in CLAUDE.md, and .env.example. Nothing outside this.

Output Format

Returns contract (≤35 lines, no raw file dumps):

High-Risk Match Logic — File: scripts/lint-plan.sh:181–222

  • Gather union: Files-in-Scope (table column 2, stripped whitespace) + per-task File: paths (with :lines suffix stripped)
  • De-dup and iterate: for each path, test against extended-regex RAD_HIGH_RISK_PATTERNS with grep -qE
  • Match = warn (never error); default pattern is auth|payment|billing|migration|secret|credential|token
  • Env var: RAD_HIGH_RISK_PATTERNS override (pipe-separated alternation); unset/empty disables

Scope-Set Computation — File: scripts/check-scope.sh:26–66

  • Files-in-Scope: table column 2 (whitespace/backtick stripped) → SCOPE_LIST
  • Tests-to-Write: each line with delimiter; extract path after (whitespace/backtick stripped) → append to scope
  • Always-allow prefixes: .agents/logs/, .agents/plans/, .agents/state/, .agents/findings.jsonl
  • Membership test: exact line match in newline-delimited list (bash 3.2+ compat)

Env-Var Doc Pattern — File: CLAUDE.md:178–195, .env.example:37–42

  • CLAUDE.md: ### heading, followed by description, then markdown code fence with VAR=pattern and inline comments
  • .env.example: hash comment line, VAR=value (commented), inline comment on next line if needed
  • Example anchor: RAD_HIGH_RISK_PATTERNS=auth|payment|billing|migration|secret|credential|token with explanation text above

Read the full file on GitHub · 51 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 · 51 lines · 53 tokens per session scan A afc00a05f3f8

Subscribe to this mod's changes

classifier-surface-mapper is an agent published in the GitHub repository seanrreid/RAD_framework (5 stars, last pushed 6d ago), licensed MIT. It adds 53 tokens to every session and 768 once invoked, about $0.0003 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.