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.
npx skills add sidiangongyuan/codex-skills-library --skill feishu-paper-readinggit clone --depth 1 https://github.com/sidiangongyuan/codex-skills-libraryWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sidiangongyuan/codex-skills-library/feishu-paper-reading)<a href="https://agentmods.dev/skills/sidiangongyuan/codex-skills-library/feishu-paper-reading"><img src="https://agentmods.dev/badge/skills/sidiangongyuan/codex-skills-library/feishu-paper-reading/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/sidiangongyuan/codex-skills-library/feishu-paper-reading"><img src="https://agentmods.dev/badge/skills/sidiangongyuan/codex-skills-library/feishu-paper-reading.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Privilege Escalation · line 256 Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.Fix: Remove references to credential paths. Use environment variables or secrets managers. For docs, use placeholder paths (e.g., /path/to/config). Never load .env or token files in production code paths.
- medium Excessive Agency · line 255 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00071 | $0.03005 |
| Opus 5 | $0.00036 | $0.01503 |
| Sonnet 5 | $0.00014 | $0.00601 |
| Haiku 4.5 | $0.00007 | $0.00300 |
Grade A, and why
feishu-paper-reading 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 11d 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.
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.
How it starts
The opening of the file, as written. The whole thing — 303 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Feishu Paper Reading
Turn a request for recent literature into an evidence-grounded reading artifact, not a ranked list of titles. Search broadly, select deliberately, read the full papers, preserve source anchors, synthesize across papers, and publish only after the report is complete.
Load The References
Read these files before acting:
references/evidence-policy.mdbefore searching or selecting papers.references/report-schema.mdbefore extracting evidence or drafting.references/feishu-publishing.mdonly when the requested destination is Feishu or another connected document surface.references/feishu-onboarding.mdwhenever Feishu is requested but a verified writable delivery path is not yet available.
Use scripts/validate_digest.py on a Markdown report before publishing when a
local runtime is available.
Use scripts/check_feishu_connection.py --json for a read-only local
discovery pass when the official Feishu CLI route may be needed; it never
executes a PATH candidate. Execution requires the installer-returned absolute
path and executable SHA-256 plus an explicit named profile and Feishu/Lark
brand, as documented in the onboarding reference.
Use the onboarding reference's installer, isolated
scripts/run_feishu_config_init.py, and protected authorization helper only
after consent. A fresh configuration must remain in the helper-returned
dedicated config/data directories; bind those exact directories, profile,
brand, and executable hash through preflight, authorization, publication, and
readback. Use scripts/publication_checkpoint.py before the first Feishu create
request. Launch every direct lark-cli child with
scripts/feishu_process_environment.py's minimal environment builder so
ambient CLI, workspace, proxy, or custom-CA selectors cannot redirect it.
Resolve The Brief
Infer harmless defaults and ask only when a missing choice would materially change the result.
Default to:
- topic: derive it from the request;
- window: previous 30 calendar days;
- count: 5 deeply read papers;
- selection: technical quality first, then observable attention;
- language: Chinese explanation with original paper titles, terminology, and short original-language evidence excerpts;
- depth: full paper plus relevant appendix, figures, and tables;
- destination: a new Feishu document, using guided least-privilege onboarding when necessary and accepted; otherwise a complete Markdown report.
What ships with it
13 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.
- agents/openai.yaml 569 B
- LICENSE 1.0 KB
- references/evidence-policy.md 4.2 KB
- references/feishu-onboarding.md 40 KB
- references/feishu-publishing.md 6.2 KB
- references/report-schema.md 4.7 KB
- scripts/check_feishu_connection.py 54 KB runs code
- scripts/feishu_process_environment.py 1.5 KB runs code
- scripts/install_lark_cli.py 38 KB runs code
- scripts/publication_checkpoint.py 41 KB runs code
- scripts/run_feishu_auth.py 80 KB runs code
- scripts/run_feishu_config_init.py 61 KB runs code
- scripts/validate_digest.py 17 KB runs code
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.
- 11d ago First seen · 303 lines · 71 tokens per session scan A 6c15d4334d94
feishu-paper-reading is a skill published in the GitHub repository sidiangongyuan/codex-skills-library (8 stars, last pushed 6d ago), licensed MIT. It adds 71 tokens to every session and 3,005 once invoked, about $0.0004 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.
Other skills, from other repositories
skills-manager-cli
Drive the Skills Manager CLI (skm) to initialize the hub, adopt unmanaged skills, list/enable/disable skills per AI tool, and doctor/fix symlink sync. Use whenever the user or an agent needs to manage skills from a terminal, SSH session, CI job, or headless machine; when a skill is missing in Claude Code, Codex…
superloopy-loop
Use Superloopy's lightweight strict-evidence loop for Codex tasks that need durable progress, criteria, and artifact-backed completion.
superloopy-frontend
Use only after explicit Codex $superloopy:superloopy-frontend or Claude Code /superloopy:superloopy-frontend invocation for supported screen-based application UI across browser-hosted Web, interactive deployed content-led Web, desktop, mobile/tablet, embedded/hybrid, Qt, custom-rendered, or mixed targets, such a task…
say-it-straight
Use only after explicit Codex $superloopy:say-it-straight or Claude Code /superloopy:say-it-straight invocation to make supplied or requested prose direct, concise, and natural without changing facts or protected text.
superloopy-research
Use only after explicit Codex $superloopy:superloopy-research or Claude Code /superloopy:superloopy-research invocation, a research task started with a leading loopy or 루피 (such as loopy research), or an active Superloopy loop explicitly routing a research deliverable here. Evidence-backed Superloopy research…
codex-delegate
Delegates implementation-heavy or repetitive coding work (batch edits, boilerplate, multi-file refactors with clear patterns, test scaffolding) from Claude to OpenAI Codex CLI. Use when token cost outweighs judgment cost. Trigger phrases include "delegate to codex", "let codex do this", "batch refactor across files"…