Skip to content

dvy1987/agent-loom / .agents/skills/apply-paper-to-project

apply-paper-to-project

Verified

Apply validated research paper insights to the current project codebase — improving architecture, code patterns, testing strategies, documentation, or workflows based on empirical findings. Load when learn-from-paper routes insights to the current project, or when the user asks to apply paper findings to this project, improve this codebase with research, use this paper to improve my project, or apply research to my code. Also triggers on "apply this to my project", "how can this paper help my codebase", "use these findings here", "implement paper recommendations", "use research to improve my code". Always called AFTER learn-from-paper has completed credibility and security checks — never ingests papers directly.

FileKindSize
SKILL.mdskill md7.8 KB
references/examples.mdmarkdown2.2 KB

references/examples.md

View on GitHub
# Apply Paper to Project — Full Worked Examples

Skill: `apply-paper-to-project` | Enriched from SKILL.md (improve-skills pass, SKIP_RESEARCH).

## Example 1 — Documented workflow

**Input:** Apply the insights from that SkillReducer paper to my current project

**Output:**
```
═══ Project Improvement Plan (from: SkillReducer) ═══

Directly Applicable:
1. compress-skill/SKILL.md — add content classification step before compression
   (paper: 60%+ of skill content is non-actionable background)
2. universal-skill-creator/SKILL.md — add SkillReducer taxonomy to Step 6
   (paper: classify blocks as CORE/WORKFLOW/BACKGROUND before writing)

Not Applicable:
- Token counting optimization — project doesn't do runtime token management

Estimated scope: 2 files, small change
Approve? (all / select / reject)
```

## Example 2 — Step-by-step execution

**Input:** "Run `apply-paper-to-project` on [concrete task]"

**Agent actions:**
1. Receive Insights
2. Understand the Project
3. Match Insights to Project
4. Present Improvement Plan
5. Apply Changes (with user approval)
6. Document Changes

**Impact Report shape:**
```
═══ Applied Research Insights ═══
Paper: [title] ([venue], [year])
Project: [project name/path]

Changes Applied:
1. [file:line] — [what changed] — [paper evidence]
2. [file:line] — [what changed] — [paper evidence]

Deferred (architecturally relevant, needs planning):
- [area] — [recommendation]

ADR created: [path or N/A]
Tests: [passed/failed/N/A]
```

## Example 3 — Gotcha application

**Input:** Task hits a non-obvious edge case

**Apply:**
- Research findings are general — the project's specific constraints may make a technique inapplicable even if the paper proves it works in theory. Always check the project context.
- Don't over-apply. If a paper finds "technique X improves performance by 15%," but the project doesn't have a performance problem, don't apply it.
- Papers often test on specific languages/frameworks. A technique proven for Python may not translate to Go or Rust idiomatically. Adapt, don't copy.
- Never add a dependency just because a paper recommends it. Check if the project already has an equivalent.

---

See `SKILL.md` for hard rules and verification checklist.
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall