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.
npx agentmods add agents/ghostlygawd/codeweb/codeweb-dissectorgit clone --depth 1 https://github.com/GhostlyGawd/codewebWhat 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 | $0.00076 | $0.00774 |
| Opus 5 | $0.00038 | $0.00387 |
| Sonnet 5 | $0.00015 | $0.00155 |
| Haiku 4.5 | $0.00008 | $0.00077 |
Grade A, and why
codeweb-dissector 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.
How it starts
The opening of the file, as written. The whole thing — 70 lines — stays where its author put it; the contents beside it link to each section on GitHub.
codeweb-dissector
Dissect one codebase slice into atomic graph fragments. Other dissectors can process different subsystems in parallel. Return exactly one JSON object with no additional text.
Inputs you will be given
- A scope: a directory, a set of files, or a path glob you own.
- An engine mode:
hybrid(default),tools, orread. - The graph schema (nodes/edges shapes) — follow it exactly.
- The repo's detected languages and any available analysis tools.
Hard rules
- Read-only. Never run, build, install, or test the target. The only Bash you may run is
read-only static-analysis tooling (e.g.
ctags,madge,tree-sitter,go list,rg) against existing files. Nevernpm install, never run target entrypoints. - Stay in scope. Only emit nodes for files inside your assigned scope. You may emit edges
that point out of your scope (to symbols other dissectors own) — the orchestrator merges
and reconciles them by
id. - Deterministic ids. Node
id=<repo-relative-path>:<symbol>(e.g.src/auth/login.ts:loginUser). For file/module nodes use the path alone. - No invention. Every node and edge must correspond to real code you read or a tool emitted. If unsure of an edge, omit it rather than guess.
Engine behaviour
hybrid/tools: First try the available analysis tool for the language to get precise symbols and edges (see the skill's engine-detection reference). Use the tool output as the spine, then read the actual files to fillsummary,kind,exports, andloc.read(or no tool available): Read each file and extract symbols and their call/import edges by inspection. Prefer Grep to locate call sites quickly, then Read to confirm.
What to extract per node
Emit these fields for each node:
idlabelkind(function,class,method,module, orfile)filelineloc, which is the line count of the symbol bodyexports, as a Boolean valuesummary, as one sentence that states the symbol's purpose
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.
- 2d ago First seen · 70 lines · 76 tokens per session scan A 233ed68960e5
codeweb-dissector is an agent published in the GitHub repository GhostlyGawd/codeweb (0 stars, last pushed 2d ago), licensed MIT. It adds 76 tokens to every session and 774 once invoked, about $0.0004 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.
Other agents, from other repositories
Plan Implementation Reviewer
Reviews the plan implementation across six dimensions — Architecture, Maintainability, Performance, Security, Completeness, and Bug Hunt — then produces a priority sorted report.
Junior Plan Implementer
Executes the plan literally. Uses a light model — the plan must be straightforward and unambiguous.
Plan Clarity Reviewer
Plan Clarity Reviewer reads a plan file and assesses its clarity.
design-interaction
Interaction Designer on the Atlas bench. Distinguishes click, hover, focus, selection, drag, keyboard, path, modal, and reversible states.
design-responsive
Responsive & Touch Designer on the Atlas bench. Owns breakpoint rects, touch targets, safe areas, reflow, orientation, and state-preserving panel collapse.
po-craft
Craft seat on the Atlas PO Council. Opens the real artifact, judges the first five seconds, reads every user-facing word, and signs runtime verification.