菜单
Use for budgeted, structure-aware previews of large or hard-to-inspect JSON/YAML, logs, code, or notebooks before deciding what to inspect next with `hson`.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Audit the structural and logical integrity of a mathematics paper — quantifier scope, domain-of-definition guards, proof/computation honesty, theorem hierarchy, contribution-list discipline (mapping every claimed contribution to its formal result, prior baseline, concrete gain, and independent significance), stale claims, notation accuracy at introduction, standing assumptions, and undefined relation symbols or coined terms in summary prose — without touching prose style or rhetorical prose inflation (structural inflation of contribution lists and result hierarchy is in scope).
Audit a LaTeX mathematics document for notation drift — symbols used under different names across sections, orphaned macros, undeclared aliases (including cross-language pairs where a term defined in one language is later used in another), symbols or relation signs used with no definition site anywhere, symbol reuse across scopes, and first-use-after-gap without back-reference — without evaluating mathematical correctness or prose style.
Review or rewrite mathematical prose so an edit, paraphrase, terminology migration, or explanatory description preserves the exact mathematical meaning fixed by its semantic source — referent identity, operation versus result, domain/codomain and representation level, quantifier and equality mode, role and provenance, notation role, and classified-before-replacement terminology migration with concept-cluster synchronization — without auditing claim-versus-proof strength (use math-claim-integrity), notation bookkeeping (use math-notation-consistency), Japanese style or translation events (use wabun-math-style), or audience prerequisites and ordering (use exposition-flow).
Detect and correct Japanese-language anti-patterns in mathematical writing — epistemic hedges weakening proved claims, incorrect verb tense, passive/active confusion in proofs, ambiguous particles, unjustified 明らか, decorative connectives, redundant meta-discourse, double negation, stacked の, generic/general conflation, borrowed non-mathematical vocabulary, vague category names for explicit formulas, full-width semicolons or untranslated English, canonical-terminology drift, and translation-level concept conflation — without evaluating mathematical correctness or structural hierarchy.
Turn repository-grounded source material into a self-contained, audience-admissible document or presentation by modelling the intended audience, extracting information units with explicit discourse roles, and enforcing dependency-ordered exposition in which every load-bearing statement follows the prerequisites needed to interpret it — without factual verification or citation checking (use sot-integrity), prose polishing (use deslop-prose), mathematical-correctness review (use the mathematical-writing skills), or typography and slide layout.
Workspace-level Git cleanup — discover repositories under the current directory and prune merged branches, merged linked worktrees, stale remote-tracking refs, and stale worktree metadata.
| name | headson |
| description | Use for budgeted, structure-aware previews of large or hard-to-inspect JSON/YAML, logs, code, or notebooks before deciding what to inspect next with `hson`. |
Use this skill when a task requires a compact preview of structured or large text data before deciding what to inspect next.
headson is the package name. The CLI command is hson.
Use hson for:
Do not use hson when exact extraction, transformation, schema validation, or deterministic data processing is required. Use jq, Python, or a purpose-built parser for those tasks.
If hson availability is unknown, check:
command -v hson
hson --version
If missing, recommend a user-controlled install. Prefer pinned Cargo when version control matters:
cargo install headson --version 0.17.0 --locked
Homebrew convenience path:
brew install headson
Do not recommend pip install headson as the primary CLI install path. The Python package provides Python bindings, not the normal hson CLI workflow expected here.
hson -c 1200 data.json
hson -c 1200 -f json -t strict data.json
hson -c 1200 -f yaml -t detailed config.yaml
Use -c, --bytes for a per-file byte budget. Use -u, --chars for a per-file character budget. Use -n, --lines for a per-file line budget.
Use -t strict when the preview itself must be machine-readable. Use -t default or -t detailed for human inspection.
hson -c 1200 events.jsonl
hson -c 1200 events.ndjson
hson -c 300 -C 2000 logs/*.json
hson --tree --glob 'src/**/*' -n 8 -N 120
Use -C, --global-bytes or -N, --global-lines when the entire output must stay within a shared budget.
Use --tree for a directory-tree view with inline previews.
hson --grep 'error|warning|traceback' -c 800 -C 3000 logs/*.json
hson --igrep 'exception|failed|timeout' -c 800 -C 3000 logs/*.json
Use strong grep when matches must remain visible under tight budgets. Matching keys, values, or lines and their ancestors are preserved before the remaining budget is spent.
Use --weak-grep or --weak-igrep only when matches should be prioritised but not guaranteed.
hson --head -c 1200 data.json
hson --tail -c 1200 data.json
Use --head to prefer early array entries. Use --tail to prefer late array entries. This affects arrays, not string truncation.
.ipynb files are JSON, but handle them explicitly rather than relying on extension auto-detection. Force JSON ingestion and rendering:
hson -i json -f json -c 2500 --string-cap 120 notebook.ipynb
Prefer early cells:
hson -i json -f json --head -c 3000 --string-cap 120 notebook.ipynb
Prefer later cells and recent outputs:
hson -i json -f json --tail -c 3000 --string-cap 120 notebook.ipynb
Keep important output and error fields visible:
hson -i json -f json \
--grep 'traceback|ename|evalue|text/plain|image/png|outputs|source' \
-c 3000 --string-cap 120 notebook.ipynb
For notebooks with very large encoded image data, keep --string-cap low enough to preserve structure without flooding the context.
--grep when the task names relevant keys, errors, warnings, or output types.--head or --tail when the beginning or end of arrays is more relevant.jq or Python once the required path or schema is known.Use hson before jq when the structure is unknown or the file is too large to inspect directly. Use jq after the relevant path or schema is known.
hson -c 1200 huge.json
jq '.items[] | {id, name}' huge.json
hson is for previewing, summarising, and deciding where to look under an explicit output budget.
jq is for exact querying, extraction, filtering, and transformation.
When reporting findings from hson, state that the output is a preview, not the full file. Mention any budget, grep, head/tail, or string-cap settings that shaped the view when they affect the conclusion.