Appendix Table Writer

v1.0.0

Curate reader-facing survey tables for the Appendix (clean layout + high information density), using only in-scope evidence and existing citation keys. **Tri...

0· 91·1 current·1 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
medium confidence
Purpose & Capability
The name/description ask for curated appendix tables and the repo contains code and pipeline docs that implement table curation from local evidence packs, anchor sheets, and citations. Required binaries (python3/python) are proportional to the task.
Instruction Scope
SKILL.md explicitly limits inputs/outputs and guardrails (no invented facts; validate citation keys; no network). The runtime script (scripts/run.py) reads workspace artifacts listed in SKILL.md (subsection_briefs.jsonl, evidence_drafts.jsonl, anchor_sheet.jsonl, citations/ref.bib) and writes outline/tables_appendix.md and a report. It does not try to read unrelated system files or environment variables in the visible code.
Install Mechanism
No install spec (instruction-only) but the skill bundle includes executable Python scripts and supporting tooling modules; the runtime expects Python on PATH. Absence of an install spec is not itself unsafe, but users should note that code shipped with the skill will be executed by the agent (no third-party downloads seen in the inspected files).
Credentials
The skill requests no environment variables, no credentials, and no config paths. Its inputs are repository/workspace files only. There are no unexpected secret requests or unrelated cloud credentials required.
Persistence & Privilege
The skill is not marked always:true and does not attempt to modify other skills or global agent settings in the inspected code. It writes output artifacts into the provided workspace paths only.
Assessment
This skill appears coherent and implements its stated purpose using Python scripts that process local evidence files and produce Markdown tables. Before installing or allowing autonomous runs: 1) Confirm your agent will pass a workspace that only contains intended inputs (avoid running against a workspace with secrets). 2) If you have low trust, run the included scripts manually in an isolated environment to inspect outputs. 3) Note there is no network access requested, but the bundle contains non-trivial tooling files (e.g., a large quality_gate module) — if you need higher assurance, skim or audit those modules for behaviors you disallow. Running in a sandbox or review by a developer is recommended if you plan to grant autonomous invocation on sensitive data.

Like a lobster shell, security has layers — review code before you run it.

latestvk970jht0kchbtncy1aew69zhrh837bm4

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Runtime requirements

Any binpython3, python

Comments