submission-venue-selector - SKILL.md Agent Skill

name: submission-venue-selector description: Use when selecting, ranking, or validating submission venues for an existing scholarly manuscript or draft venue shortlist. A deliverable venue recommendation requires comparator-paper evidence for every ranked venue; bibliography overlap and offline placeholders are discovery signals only. Do not use for generic draft review, rewriting, paper retrieval, paper download, Zotero mutation, or one-off venue facts. metadata: short-description: Evidence-gated journal and conference venue selection

Submission Venue Selector

Use this skill to build an evidence-backed venue dossier for a scholarly draft. It can run a runtime helper that extracts references, builds candidate venues, collects or records related-paper evidence, scores fit, and writes journal recommendations with heuristic acceptance-chance intervals.

No Shallow Shortlist

A venue recommendation is deliverable only when each ranked venue has comparator-paper evidence. Bibliography overlap, venue reputation, and offline placeholders are discovery signals, not comparator-paper evidence. If this evidence is unavailable, output incomplete analysis, keep delivery status not-ready, and do not present a final ranked shortlist.

The final report must still list every candidate journal with an estimated acceptance-chance interval, confidence, calculation class, modifier breakdown, and caveats. These estimates are heuristics, not predictions or guarantees. Before writing a deliverable report, load writing-style-settings.md and record the active style profile. For mathematical, TCS, graph-theoretic, Lean, or LaTeX manuscripts, also load math-manuscript-style.md.

Routing Boundary

Use this skill for:

"Where should I submit this draft?"
journal or conference shortlist ranking
validating an existing venue shortlist
comparing venue fit against a draft and its bibliography

Do not use this skill for:

draft rewriting or polishing; use draft-writing
paper retrieval or downloads; use zotero and then getscipapers-requester
general paper review; use paper-review or agent-group-discuss
one-off venue facts that do not require a draft-fit dossier

Runtime Helper

POSIX:

bash "$AAS_RUNTIME_ROOT/run_skill.sh" \
  skills/submission-venue-selector/run_submission_venue_selector.sh \
  run --dir /path/to/venue-run --draft /path/to/draft.tex --offline

This offline command is a smoke/provisional run. It is not a deliverable venue recommendation unless trusted fixture/cache comparator evidence is provided.

Windows PowerShell:

$runtime = if ($env:AAS_RUNTIME_ROOT) { $env:AAS_RUNTIME_ROOT } else { "$env:LOCALAPPDATA\ai-agents-skills\runtime" }
& "$runtime\run_skill.ps1" "skills/submission-venue-selector/run_submission_venue_selector.ps1" run --dir "$env:USERPROFILE\venue-run" --draft "$env:USERPROFILE\drafts\paper.tex" --offline

Windows CMD:

"%AAS_RUNTIME_ROOT%\run_skill.bat" skills/submission-venue-selector/run_submission_venue_selector.bat run --dir "%USERPROFILE%\venue-run" --draft "%USERPROFILE%\drafts\paper.tex" --offline

Useful commands:

init --dir <workspace> --draft <path>
plan --dir <workspace>
extract --dir <workspace>
privacy-gate --dir <workspace>
providers --check --dir <workspace>
resolve --dir <workspace> [--allow-network --allow-provider <name>]
expand --dir <workspace>
venues --dir <workspace>
recent --dir <workspace>
score --dir <workspace>
report --dir <workspace>
validate --dir <workspace>
run --dir <workspace> --draft <path> --offline for smoke/provisional output
purge --dir <workspace>
smoke

Safety Defaults

Local/offline by default.
Network requires a prior ok privacy-gate, --allow-network, and explicit --allow-provider <name>.
Downloads, Zotero mutations, and Unpaywall email use are forbidden unless explicitly enabled by command flags.
Raw draft text is not persisted unless --retain-draft-text is used.
Reports separate observed evidence from inferred venue fit and must mark placeholder-only output as incomplete analysis.
Acceptance-chance estimates are required in reports, but bare percentages and predictive acceptance claims are invalid.
Final report artifacts should record style_profile_ref, active_overlays, active_requirement_ids, and style_applied. A bare style_applied: true value is not enough unless the artifact or workflow ledger also records the loaded policy and selected requirements.

Workflow

Start with init or run.
Confirm privacy-gate before any live provider calls.
Use providers --check to record available provider capabilities.
Resolve references and derive candidate venues.
Score venues with criterion-level evidence IDs and acceptance-chance intervals.
Run validate before treating the recommendation as deliverable.

Read When Needed

references/provider-policy.md: provider capabilities and network rules.
references/artifact-schema.md: artifact files, IDs, and validation rules.
references/scoring-rubric.md: scoring criteria and delivery statuses.
references/report-contract.md: recommendation report structure.
references/privacy-and-network-policy.md: draft privacy, cache, and mutation boundaries.