Skip to content

simiancraft/simiancraft-skills / skills/appraise-github-issues

appraise-github-issues

Verified

Appraise a GitHub issue backlog unattended: for each issue, decide whether it is still real and how big it is, then apply the answer to the tracker as a size label, a needs-decision or needs-human label with the question stated, or a close carrying a re-checkable receipt. A close is confirmed by a second engine before it is made. Runs once, on one issue, over the whole backlog, or as a periodic heartbeat, and is the sizing stage of burn-down-github-issues. Use when the task is "size the backlog", "appraise every issue in this repo", "close what is already done", "triage recent issues without fixing them", or "keep the backlog sized". Requires a per-repository config (its own, or the burndown's) and the codex and claude CLIs for the default seats. Skip for fixing anything; this skill never writes to the repository.

FileKindSize
SKILL.mdskill md7.1 KB
appraise.tsscript12.7 KB
lib/appraise.test.tsscript2.8 KB
lib/appraise.tsscript30.9 KB
lib/callbacks.tsscript4.1 KB
prompts/appraise.mdmarkdown6.2 KB
prompts/confirm-close.mdmarkdown3.3 KB
references/adopting.mdmarkdown4.0 KB
references/callbacks.mdmarkdown3.4 KB

references/adopting.md

View on GitHub
# Adopting appraise-github-issues

The command reads one config file at the repository root. A repository that has already adopted
`burn-down-github-issues` needs nothing more: `appraise.ts` reads `burn-down-github-issues.config.ts`
when no `appraise-github-issues.config.ts` exists, and the burndown's `ageDays`, `appraiseLimit`,
`appraiserConcurrency`, `skipLabels`, and `seats.appraiser` are the same knobs.

A repository adopting appraisal alone writes `appraise-github-issues.config.ts`:

```ts
export default {
  project: {
    // The same `project` shape as burn-down-github-issues; only these fields are read here.
    name: 'YourApp',
    repo: 'your-org/your-app',
    remote: 'origin',
    baseBranch: 'main',
    conventionDocs: ['CONTRIBUTING.md'],   // what an issue's prescribed remedy is read against
    sizingScale: 'docs/sizing.md',          // where the point scale is written
    worktreeRoot: '../.your-app-loop',      // runs/ and scratch directories live under it
    // The loader validates the whole project shape, so the remaining fields must be present
    // even though appraisal never uses them; copy them from the burndown template.
    evidenceBranch: '__evidence_locker__',
    checkCommand: 'bun run verify',
    installCommand: 'bun install --frozen-lockfile',
    sharedServices: [],
    portBase: 41000,
    portSpan: 1000,
    pathAliases: [{ prefix: '@/', dir: 'src' }],
    sourceExtensions: ['.ts', '.tsx'],
    alwaysInvalidates: [],
    touchPaths: { migration: [], ci: [] },
  },
  ageDays: 30,                 // only issues opened this recently, unless --all; --age-days <n> overrides it for one run
  appraiseLimit: 12,           // per run, unless --limit
  appraiserConcurrency: 3,
  confirmCloses: true,         // a second engine must agree before an issue closes
  maxAppraiseAttempts: 3,      // failed appraisals an issue absorbs before loop/dlq: appraisal
  sizeCallbackTimeoutMinutes: 0, // how long a size callback may run; 0 means no timer
  skipLabels: ['needs-decision', 'needs-human', 'loop/skip', 'loop/parked'],
  callbacksDir: '<worktreeRoot>/appraisal-callbacks', // where on-size-<N> callbacks live; see references/callbacks.md
  seats: { appraiser: 'codex:gpt-5.6-sol', confirmer: 'claude:claude-opus-5', callback: 'codex:gpt-5.6-sol' },
};
```

## What it writes to the tracker

Labels it creates on start, idempotently: `needs-decision`, `needs-human`, `loop/skip`,
`loop/parked`, any `loop/dlq: <phase>`. Labels it applies: `size: N`, `needs-decision`, `needs-human`. Comments:
the question for a hand-off, the receipt for a close, both opinions for a disputed close, and the
reason for a re-size. Closes: only `already-fixed` and `obsolete`, and only after the confirmer
agrees unless `confirmCloses` is off.

An issue carrying `needs-decision`, `needs-human`, `loop/skip`, `loop/parked`, or a `loop/dlq: <phase>` is
never selected; a person removes the label to put it back in reach. `--issue <n>` bypasses the
window and the size filter but not those labels.

## First-run order

1. `appraise.ts --dry-run`: prints what it would select; no lock is taken, no agent runs, and nothing but `runs/appraise.log` is written.
2. `appraise.ts --issue <n>` on one issue you already know the answer for. Read the comment or the
   label it left and check that a stranger could re-check it.
3. `appraise.ts --limit 3`, then a full window.
4. `appraise.ts --all` once, if the backlog predates the window. Expect the closes to be the
   highest-yield part; read a few of the receipts before trusting the rest.
5. `appraise.ts --every <minutes>` in a terminal you can watch, if you want the backlog kept sized
   between burndown runs.

## Boundaries

`--all --include-sized` re-judges every open issue and will comment on every sizing disagreement it
finds. On a large backlog that is a lot of comments; run it once, deliberately, not on a cadence.

`--no-confirm` removes the only independent check on a close. Use it for a tracker you would let
one model close issues on unsupervised, and nowhere else.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall