paper-orchestra

Name: paper-orchestra
Author: woodfishhhh

31stars1forks

How to install paper-orchestra?

woodfishhhh/ez_math_model/paper-orchestra

$npx -y skills add woodfishhhh/EZ_math_model --skill paper-orchestra

Installs into the current project.

Summary

Orchestrate the full PaperOrchestra (Song et al., 2026, arXiv:2604.05018) five-agent pipeline to turn unstructured research materials (idea, experimental log, LaTeX template, conference guidelines, optional figures) into a submission-ready LaTeX manuscript and compiled PDF. TRIGG

Files · 1

View on GitHub

SKILL.md

1	# paper-orchestra (Orchestrator)
2
3	Top-level driver for the PaperOrchestra pipeline. Read this document and follow
4	the steps below. The detailed prompts and rules live in each sub-skill's
5	`SKILL.md` and `references/` directories — you (the host agent) will load them
6	as you go.
7
8	> Source paper: Song et al., *PaperOrchestra: A Multi-Agent Framework for
9	> Automated AI Research Paper Writing*, arXiv:2604.05018, 2026.
10	> <https://arxiv.org/pdf/2604.05018>
11
12	## What this skill produces
13
14	A complete submission package `P = (paper.tex, paper.pdf)` written into
15	`workspace/final/`, plus a full audit trail under `workspace/` (outline,
16	figures, refs, drafts, refinement worklog, provenance snapshot).
17
18	## Inputs (the (I, E, T, G, F) tuple from the paper)
19
20	The workspace MUST contain:
21
22	\| File \| Symbol \| Required \| Description \|
23	\|---\|---\|---\|---\|
24	\| `workspace/inputs/idea.md` \| `I` \| yes \| Idea Summary (Sparse or Dense variant — see `references/io-contract.md`) \|
25	\| `workspace/inputs/experimental_log.md` \| `E` \| yes \| Experimental Log: setup, raw numeric data, qualitative observations \|
26	\| `workspace/inputs/template.tex` \| `T` \| yes \| LaTeX template for the target conference (with `\section{...}` commands) \|
27	\| `workspace/inputs/conference_guidelines.md` \| `G` \| yes \| Formatting rules, page limit, mandatory sections \|
28	\| `workspace/inputs/figures/` \| `F` \| no \| Optional pre-existing figures. If empty, the plotting agent generates everything. \|
29
30	`scripts/init_workspace.py` will scaffold this layout. `scripts/validate_inputs.py`
31	will check it before the pipeline runs.
32
33	## Pipeline (read `references/pipeline.md` for the full diagram)
34
35	```
36	Step 1: Outline ──▶ outline.json (1 LLM call)
37	Step 2: Plotting ─┐
38	├──▶ figures/*.png + captions.json (~20-30 calls)
39	Step 3: Lit Review ─┘ (~20-30 calls)
40	intro_relwork.tex + refs.bib
41
42	Step 4: Section Writing ──▶ drafts/paper.tex (1 LLM call)
43	Step 5: Content Refine ──▶ final/paper.tex + final/paper.pdf (~5-7 calls, ~3 iters)
44	```
45
46	Step 2 and Step 3 are independent and MUST run in parallel when your host
47	supports parallel sub-agents. If not, run Step 3 first (it has the longer wall
48	time due to Semantic Scholar rate limits) and Step 2 second.
49
50	## Critical pre-instruction (read once, apply always)
51
52	Before any LLM call that writes paper content (outline, intro/related work,
53	section writing, refinement), you MUST prepend the Anti-Leakage Prompt at
54	`references/anti-leakage-prompt.md` to your system prompt. This is verbatim
55	from Appendix D.4 of the paper and prevents pre-training-data leakage. The
56	paper applies it uniformly across all baselines for fair comparison; we apply
57	it for fidelity and to keep generated papers grounded in the user's inputs.
58
59	## Step-by-step execution
60
61	### 0. Pre-flight Checks
62
63	Before running the pipeline, perform the following quality gates in order:
64
65	```bash
66	# 1. Scaffold the workspace
67	python skills/paper-orchestra/scripts/init_workspace.py --out workspace/
68	# user drops their inputs into workspace/inputs/
69
70	# 2. Validate required files are present and well-formed
71	python skills/paper-orchestra/scripts/validate_inputs.py --workspace workspace/
72
73	# 3. Check input density — idea and experimental log must meet minimum thresholds
74	python skills/paper-orchestra/scripts/check_idea_density.py \
75	--idea workspace/inputs/idea.md \
76	--log workspace/inputs/experimental_log.md
77
78	# 4. Cross-validate consistency between idea and experimental log
79	python skills/paper-orchestra/scripts/validate_consistency.py \
80	--idea workspace/inputs/idea.md \
81	--log workspace/inputs/experimental_log.md
82	```
83
84	If `validate_inputs.py` or `check_idea_density.py` fail (exit code 1 or 2), stop
85	and tell the user what's missing or below threshold — do not proceed until fixed.
86
87	`validate_consistency.py` produces warnings only (exit code 1 = WARN, non-blocking);
88	report warnings to the user but continue.
89
90	Before failing on missing inputs, check whether aggregation can supply them:
91
92	\| Inputs state \| Action \|
93	\|---\|---\|
94	\| `idea.md` and `experimental_log.md` both present and non-empty \| Continue to Step 1. \|
95	\| Either is missing/empty, and the user men

People who liked this also viewed

6 picks

Type

paper-orchestra

31stars1forks

View on GitHub

How to install paper-orchestra?

woodfishhhh/ez_math_model/paper-orchestra

$npx -y skills add woodfishhhh/EZ_math_model --skill paper-orchestra

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# paper-orchestra (Orchestrator)
2
3	Top-level driver for the PaperOrchestra pipeline. Read this document and follow
4	the steps below. The detailed prompts and rules live in each sub-skill's
5	`SKILL.md` and `references/` directories — you (the host agent) will load them
6	as you go.
7
8	> Source paper: Song et al., *PaperOrchestra: A Multi-Agent Framework for
9	> Automated AI Research Paper Writing*, arXiv:2604.05018, 2026.
10	> <https://arxiv.org/pdf/2604.05018>
11
12	## What this skill produces
13
14	A complete submission package `P = (paper.tex, paper.pdf)` written into
15	`workspace/final/`, plus a full audit trail under `workspace/` (outline,
16	figures, refs, drafts, refinement worklog, provenance snapshot).
17
18	## Inputs (the (I, E, T, G, F) tuple from the paper)
19
20	The workspace MUST contain:
21
22	\| File \| Symbol \| Required \| Description \|
23	\|---\|---\|---\|---\|
24	\| `workspace/inputs/idea.md` \| `I` \| yes \| Idea Summary (Sparse or Dense variant — see `references/io-contract.md`) \|
25	\| `workspace/inputs/experimental_log.md` \| `E` \| yes \| Experimental Log: setup, raw numeric data, qualitative observations \|
26	\| `workspace/inputs/template.tex` \| `T` \| yes \| LaTeX template for the target conference (with `\section{...}` commands) \|
27	\| `workspace/inputs/conference_guidelines.md` \| `G` \| yes \| Formatting rules, page limit, mandatory sections \|
28	\| `workspace/inputs/figures/` \| `F` \| no \| Optional pre-existing figures. If empty, the plotting agent generates everything. \|
29
30	`scripts/init_workspace.py` will scaffold this layout. `scripts/validate_inputs.py`
31	will check it before the pipeline runs.
32
33	## Pipeline (read `references/pipeline.md` for the full diagram)
34
35	```
36	Step 1: Outline ──▶ outline.json (1 LLM call)
37	Step 2: Plotting ─┐
38	├──▶ figures/*.png + captions.json (~20-30 calls)
39	Step 3: Lit Review ─┘ (~20-30 calls)
40	intro_relwork.tex + refs.bib
41
42	Step 4: Section Writing ──▶ drafts/paper.tex (1 LLM call)
43	Step 5: Content Refine ──▶ final/paper.tex + final/paper.pdf (~5-7 calls, ~3 iters)
44	```
45
46	Step 2 and Step 3 are independent and MUST run in parallel when your host
47	supports parallel sub-agents. If not, run Step 3 first (it has the longer wall
48	time due to Semantic Scholar rate limits) and Step 2 second.
49
50	## Critical pre-instruction (read once, apply always)
51
52	Before any LLM call that writes paper content (outline, intro/related work,
53	section writing, refinement), you MUST prepend the Anti-Leakage Prompt at
54	`references/anti-leakage-prompt.md` to your system prompt. This is verbatim
55	from Appendix D.4 of the paper and prevents pre-training-data leakage. The
56	paper applies it uniformly across all baselines for fair comparison; we apply
57	it for fidelity and to keep generated papers grounded in the user's inputs.
58
59	## Step-by-step execution
60
61	### 0. Pre-flight Checks
62
63	Before running the pipeline, perform the following quality gates in order:
64
65	```bash
66	# 1. Scaffold the workspace
67	python skills/paper-orchestra/scripts/init_workspace.py --out workspace/
68	# user drops their inputs into workspace/inputs/
69
70	# 2. Validate required files are present and well-formed
71	python skills/paper-orchestra/scripts/validate_inputs.py --workspace workspace/
72
73	# 3. Check input density — idea and experimental log must meet minimum thresholds
74	python skills/paper-orchestra/scripts/check_idea_density.py \
75	--idea workspace/inputs/idea.md \
76	--log workspace/inputs/experimental_log.md
77
78	# 4. Cross-validate consistency between idea and experimental log
79	python skills/paper-orchestra/scripts/validate_consistency.py \
80	--idea workspace/inputs/idea.md \
81	--log workspace/inputs/experimental_log.md
82	```
83
84	If `validate_inputs.py` or `check_idea_density.py` fail (exit code 1 or 2), stop
85	and tell the user what's missing or below threshold — do not proceed until fixed.
86
87	`validate_consistency.py` produces warnings only (exit code 1 = WARN, non-blocking);
88	report warnings to the user but continue.
89
90	Before failing on missing inputs, check whether aggregation can supply them:
91
92	\| Inputs state \| Action \|
93	\|---\|---\|
94	\| `idea.md` and `experimental_log.md` both present and non-empty \| Continue to Step 1. \|
95	\| Either is missing/empty, and the user men

People who liked this also viewed

6 picks

Type