fix-parser-gap

A workflow for adding a missing feature to an Office-file parser using test-driven development, or TDD: write a failing test, implement the feature, then make the test pass and measure the result.

In plain words
What is it for?
Use it to close gaps in parsing DOCX, XLSX, or PPTX files according to the ECMA-376 Office-file specification, then rerun benchmarks.
Why use it?
It prevents parser changes from being made without a test and checks whether the improvement works on real evaluation cases.

Skill for Claude CodeCodex

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 skills/sunholo-data/ailang-parse/fix-parser-gap
Any agent
npx skills add sunholo-data/ailang-parse --skill fix-parser-gap
Clone the repo
git clone --depth 1 https://github.com/sunholo-data/ailang-parse

Made for: Claude Code, Codex.

Per session 116 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,580 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 $0.00116 $0.01580
Opus 5 $0.00058 $0.00790
Sonnet 5 $0.00023 $0.00316
Haiku 4.5 $0.00012 $0.00158

Measured 2d ago against content hash 550bc0101067, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

fix-parser-gap 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.

The scan reads SKILL.md. This mod also ships 1 executable file (scripts/run_gap_check.sh), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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/skills/fix-parser-gap/SKILL.md · 154 lines

How it starts

The opening of the file, as written. The whole thing — 154 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Fix Parser Gap

Implements missing ECMA-376 Office parsing features using a TDD red-to-green workflow: write a failing test first, read the spec, implement in AILANG, verify the test goes green, then re-run benchmarks.

Workflow: Red -> Green -> Benchmark

Step 1: Select the Gap

If the user specifies a feature, use that. Otherwise pick the highest-priority unresolved gap from the audit:

# Show current gap scores (0% = complete gap, 100% = fully handled)
uv run benchmarks/office/eval_gaps.py --verbose

Consult resources/reference.md for the priority table and spec page numbers.

Step 2: RED — Write or Update the Failing Test

This step is critical. No implementation without a failing test first.

Two places tests live:

  1. benchmarks/office/eval_gaps.py — Gap analysis checks. Each check is a function like check_custom_heading_styles(output) that returns a score from 0.0-1.0. If a check doesn't exist for the gap, add one. If it exists but is too lenient, tighten it.

  2. benchmarks/officedocbench/ground_truth/challenge_*.json — Hand-verified ground truth for challenge files. These files have "verified": true so annotate.py won't overwrite them. If the ground truth doesn't reflect the expected correct behavior, fix it.

Also create or verify the challenge test file in data/test_files/challenge/ — these are generated by benchmarks/create_challenge_files.py using python-docx/openpyxl. If the existing challenge file doesn't test the feature well enough, update the generator.

Run the test to confirm it fails (RED):

uv run benchmarks/office/eval_gaps.py --verbose  # Check the specific gap shows 0%

Step 3: Read the Spec

Read the relevant ECMA-376 spec section for the feature being implemented:

# Spec PDF is at:
specs/ecma-376/part1/ECMA-376-1-5th-edition-december-2016-Part1.pdf

Use the page numbers from resources/reference.md. Focus on:

  • The XML element structure (tag names, attributes, child elements)
  • Where the element appears in the ZIP (which XML file)
  • How relationships work (if the feature uses r:id references)

Read the full file on GitHub · 154 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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 · 154 lines · 116 tokens per session scan A 550bc0101067

Subscribe to this mod's changes

fix-parser-gap is a skill published in the GitHub repository sunholo-data/ailang-parse (0 stars, last pushed 2d ago), licensed Apache-2.0. It adds 116 tokens to every session and 1,580 once invoked, about $0.0006 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.

Related

Other skills, from other repositories