harvest-errors

A command that reads the latest Test262 results from the js2wasm-baselines repository, matches error patterns against existing issues, and creates issues for patterns not already tracked.

In plain words
What is it for?
Use it after a Test262 run to collect the full results, identify recurring unaddressed failures, and open tracking issues for them.
Why use it?
It prevents incomplete or stale test data from hiding failures and reduces the manual work of checking whether each error already has an issue.

Command 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 commands/loopdive/js2/harvest-errors
Clone the repo
git clone --depth 1 https://github.com/loopdive/js2

Made for: Claude Code.

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,531 The whole file, excluding the scripts and references it only reads on demand.
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 $0.00000 $0.02531
Opus 5 $0.00000 $0.01265
Sonnet 5 $0.00000 $0.00506
Haiku 4.5 $0.00000 $0.00253

Measured 2d ago against content hash dbeb5397c409, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

harvest-errors 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 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

curl -fsSL https://raw.githubusercontent.com/loopdive/js2wasm-baselines/main/test262-standalone-current.jsonl \
.claude/commands/harvest-errors.md · 176 lines

How it starts

The opening of the file, as written. The whole thing — 176 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Harvest Test262 Errors

Analyze the latest test262 run results, cross-reference with existing issues, and create new issues for unaddressed error patterns.

Data source — the loopdive/js2wasm-baselines repo (authoritative)

Always harvest the full detailed run results published in loopdive/js2wasm-baselines, not any copy committed into the main repo. CI (promote-baseline in test262-sharded.yml) pushes the complete per-test results there on every merge to main. The main repo no longer carries the JSONL blob (#1528); any local benchmarks/results/*.jsonl is a trimmed, possibly-stale mirror — using it under-reports the real pass rate (it read 61.5% on 2026-06-03 when the baselines repo had 70.7%). Fetch fresh every run.

Baselines-repo file set (root of the repo, branch main):

Default (JS-host) lane Standalone lane
Full results (one JSON/test) test262-current.jsonl (~36 MB) test262-standalone-current.jsonl (~53 MB)
Latest raw run test262-results.jsonl test262-standalone-results.jsonl
Summary counts test262-current.json test262-standalone-current.json
Report rollup test262-report.json test262-standalone-report.json
Trend history runs/index.json (shared)

There are two independent test262 lanes — harvest each separately and never mix their counts (they are distinct conformance metrics on different targets):

Lane Target flags Goal tag
Default (JS-host) gc target, host imports allowed (default)
Standalone --target standalone --no-host-imports, nativeStrings goal: standalone-mode

The standalone lane measures pure-Wasm conformance (no JS runtime). Its failures are dominated by host-import leaks — features that silently fall back to a JS host import in the default lane but are refused in standalone mode. The standalone lane's dominant signal is the #NNNN citation embedded in each refusal's error string (see step 2). NOTE (verified 2026-08-01): the host_import_leak_class field appears in BOTH lanes and is actually far more common on the DEFAULT lane (41,276 records, single value dynamic_object_property) than on standalone (2,679 records, four values) — an earlier revision of this doc claimed the reverse. Do not use that field to tell the lanes apart.

Read the full file on GitHub · 176 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 · 176 lines · 0 tokens per session scan A dbeb5397c409

Subscribe to this mod's changes

harvest-errors is a command published in the GitHub repository loopdive/js2 (59 stars, last pushed 2d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 2,531 tokens. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.