issueforge CLAUDE.md

issueforge CLAUDE.md is an instructions file for coding agents from vietnamesekid/issueforge. It costs 2,950 tokens per session, scanned A, original, Apache-2.0.

Project instructions for IssueForge, a local supervisor that lets a self-hosted runner start an installed coding agent for a GitHub issue.

In plain words
What is it for?
They are for contributing code to IssueForge, understanding its issue-driven workflow, and running the project's checks correctly.
Why use it?
They keep changes within the project's boundaries and identify the required type-checking, linting, build, and test commands.

Instructions file

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 instructions/vietnamesekid/issueforge/claude-md
Clone the repo
git clone --depth 1 https://github.com/vietnamesekid/issueforge

Wrote 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.

agentmods badge for issueforge CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/vietnamesekid/issueforge/claude-md.svg)](https://agentmods.dev/instructions/vietnamesekid/issueforge/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/vietnamesekid/issueforge/claude-md"><img src="https://agentmods.dev/badge/instructions/vietnamesekid/issueforge/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 2,950 This file is loaded in full into every session.
When invoked 2,950 The same file — it is already loaded in full.
Security scan A 1 finding. 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 $0.02950 $0.02950
Opus 5 $0.01475 $0.01475
Sonnet 5 $0.00590 $0.00590
Haiku 4.5 $0.00295 $0.00295

Measured 4d ago against content hash 84e1954f821e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

issueforge CLAUDE.md scanned grade A with 1 finding 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 4d 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.

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

**Write the test first and watch it fail.** A test written after a fix used `spawnSync`
CLAUDE.md · 202 lines

How it starts

The opening of the file, as written. The whole thing — 202 lines — stays where its author put it; the contents beside it link to each section on GitHub.

IssueForge

A local-first GitHub IssueOps supervisor. A maintainer labels an issue, a self-hosted runner spawns an already-installed coding agent (Claude Code / Codex) in a SHA-pinned worktree, and the agent reports its own findings back to the issue with gh.

IssueForge is not a coding agent. It never plans steps, chooses tools, or calls a model, and it writes no GitHub content — no comments, no labels, no PRs; the harness does all of it. If a change requires planning, tool choice, or a model call, that is scope drift: say so instead of building it.

Commands

pnpm check                  # typecheck + lint + test — the arbiter, run before declaring done
pnpm test                   # runs `pnpm build` first, deliberately (see below)
pnpm test:unit              # vitest only, no build — fast loop
npx vitest run path/to.test.ts   # a single file
node apps/cli/src/main.ts --help  # run the CLI with no build step

pnpm test builds first because sqlite-run-store.test.ts spawns a second real process that imports packages/adapters/dist/index.js; without a build it warns and skips, silently losing the only two-process test. Everything else resolves @issueforge/* to src.

The one rule enforced mechanically

contracts  ->  core  ->  adapters  ->  apps/cli
 (no deps)    (pure)     (io only)     (wiring)

core imports only contracts — no node:fs, node:sqlite, execa, gh. No adapter imports @octokit or any GitHub API client. Nothing imports the CLI. tests/import-boundaries.test.ts fails CI on a violation. It is what stops IssueForge drifting into a coding-agent framework.

That test once guarded three deleted directories, caught its own ENOENT, and passed by checking nothing. When you add a boundary test, make it fail first.

Where new code goes

  • New task (fix, verify) → a TaskDefinition value (kind, buildCard, resultSchema) in adapters/src/runner/task-runner.ts, then register it in TASKS in main.ts and in INTENT_LABELS in contracts/src/github.ts. Do not copy TaskRunner — its ordering is crash-safety logic and a second copy will rot.
  • New harnessadapters/src/harness/<name>/.
  • New CLI commandapps/cli/src/commands/<name>.ts exporting a doer (runX, returns data) and a renderer (renderX, returns a string), then wire it into buildProgram() in main.ts. Commands never touch stdout themselves — that split is what makes them testable. Get dependencies from createContext(); never construct adapters directly. Set process.exitCode, never call process.exit.
  • New I/O → adapters; new rules about what a run may do → core/domain; new vocabulary → contracts.

Read the full file on GitHub · 202 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. 4d ago First seen · 202 lines · 2,950 tokens per session scan A 84e1954f821e

Subscribe to this mod's changes

issueforge CLAUDE.md is an instructions file published in the GitHub repository vietnamesekid/issueforge (2 stars, last pushed 2d ago), licensed Apache-2.0. It adds 2,950 tokens to every session, about $0.0147 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.