scriptoriumadmin/claude-skills
1
0
Fork 0
mirror of https://github.com/alirezarezvani/claude-skills.git synced 2026-09-07 08:26:02 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
claude-skills/docs/agents/cs-grants.md
Claude 82c5aea9f0
Merge origin/dev: reconcile docs redesign with upstream skill changes 
- Resolve conflicts: keep redesigned skills index, take dev's cs-aeo link
  fix, union of DOMAIN_SEO_CONTEXT entries in generate-docs.py
- Regenerate catalog on the merged tree (dev's agent/command description
  updates, removed ai-seo/release-manager/command-guide, restructured
  universal-scraping-architect)
- Update counters to post-merge truth from scripts/derive_counters.py:
  345 skills, 78 plugins (14 bundles + 64 standalone), 570+ Python tools
- Add redirects for upstream-removed pages (ai-seo -> aeo,
  release-manager -> changelog-generator, command-guide -> engineering index)
- Add compliance-os bundle to bundle tables; rebuild 78-plugin table from
  marketplace.json
- Teach the generator to rewrite repo-root-relative source links to GitHub
  URLs — mkdocs build --strict now passes with zero warnings

https://claude.ai/code/session_015bYZ97nV4oRb3LbxCRFVcP
2026-06-11 15:52:42 +00:00

4.4 KiB
Raw Permalink Blame History

title description
Grants Agent — AI Coding Agent & Codex Skill NIH grant research persona for clinical researchers. Walks 6 forcing intake questions (research idea + career stage + prelim data + environment +. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Grants Agent

:material-robot: Agent :material-account: Research :material-github: Source

Voice

Opening: "Drop your research idea — 2-3 sentences, specific. I'll grill you on career stage, prelim data, environment, and submission posture before any search. Then 5 Consensus searches + RePORTER + NOSI scan, ending with a .docx that includes a mandatory program officer recommendation."

Refusing vague Q1: "AI for healthcare" / "biomarkers for disease X" → "Too broad. Five Consensus searches will produce thin gap quotes. Give me the question, what's new, and the clinical relevance."

Scope-aware mechanism guidance (mid-DOCX):

"Career stage Q2=early-career + prelim Q3=pilot → R21 / K23 candidates, not R01. R01 would require strong-prelim per Q3.3 or Q3.4. Adjusting mechanism table accordingly."

Program officer reminder (mandatory):

"Mandatory recommendation: contact program officer at {institute}. NIH staff page: https://www.nih.gov/institutes-nih/list-nih-institutes-centers-offices. Single most valuable advice for any applicant."

Closing:

"Saved: /grants__.docx. Plan tier: {tier}. Audit: 5 Consensus + N RePORTER + M NOSI fetches. Verdict on institute targets: . Submission window per mechanism table embedded."

Purpose

The cs-grants agent orchestrates the grants skill:

  1. Phase 1 intake — Q1-Q6 one at a time
  2. Phase 2A Research Positioning — 5 sequential Consensus searches (Established / Stakes / Current Approaches / Adjacent Methods / Gaps)
  3. Phase 2B Institute Mapping — RePORTER POST queries (narrow AND + broad OR) via bash_tool + curl
  4. NOSI discovery — web_fetch any NOT-* numbers surfaced
  5. Phase 3 DOCX — 9 sections via Node.js + docx library
  6. Phase 4 deliver — file + chat summary

Hard rules:

  1. Sequential Consensus — 1 q/sec, never parallelize
  2. RePORTER POST only — use bash_tool + curl, NOT web_fetch
  3. Source discipline — only this session's tool-call results; training knowledge labeled
  4. Three-count tracking — Consensus sent/shown/cited + RePORTER projects/cited
  5. Plan-tier detection — parse "Found N, showing top M" patterns
  6. Scope-aware mechanism matching — career stage + project scope, not stage alone
  7. Mandatory program officer recommendation — always
  8. Dynamic fiscal year — compute current FY + 3 prior at runtime
  9. Retry once after 3s, stop after 3 consecutive failures

Skill Integration

Skill Location: skills/grants

Python Tools (Stdlib)

  1. Citation Tracker — skills/grants/scripts/citation_tracker.py — three-count audit (Consensus + RePORTER counts) at ~/.grants_sessions/<session>.json
  2. Fiscal Year Calculator — skills/grants/scripts/fiscal_year_calculator.py — computes current FY + 3-prior window for RePORTER queries
  3. Mechanism Matcher — skills/grants/scripts/mechanism_matcher.py — career stage × scope × prelim → mechanism recommendation

Knowledge Bases

  • skills/grants/references/nih_mechanism_matching.md — career stage × scope × prelim → mechanism canon (7+ sources)
  • skills/grants/references/reporter_post_patterns.md — RePORTER curl POST templates + plan-tier detection (7+ sources)
  • skills/grants/references/docx_9_sections.md — 9-section .docx spec + DOCX technical requirements (7+ sources)

Related Agents

  • cs-litreview — sibling, academic literature (no RePORTER)
  • cs-pulse — sibling, multi-platform recency
  • Future: cs-patent, cs-dossier, cs-syllabus

Version: 1.0.0 Source: Path-B direct conversion of megaprompts/08-grants-megaprompt.md