Academic Pipeline
ReviewAudited by ClawScan on May 16, 2026.
Overview
This is a coherent instruction-only academic workflow skill, with disclosed but important privacy considerations around delegated agents, persistent memory, and optional cross-model review.
Before installing, make sure you are comfortable sending the project through delegated academic sub-skills and AI-provider workflows. For confidential or unpublished work, review Hermes session storage, Hindsight Memory retention, and any optional cross-model setting before running the pipeline.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Your topic, drafts, and research materials may be passed to other skills, and those skills may use their own permitted tools.
The skill does not execute code itself, but it delegates substantial work to other skills whose permissions are outside this artifact.
This pipeline orchestrates multiple skills (research → writing → review) via `delegate_task`... sub-skills ... independently control their own tool access.
Review the permissions and behavior of the delegated research, writing, and reviewer skills before using this on confidential work.
Drafts, stage outputs, or summaries may remain available in session or long-term memory after the immediate task.
The skill explicitly stores pipeline state and key academic outputs persistently, which is coherent for resume support but can retain sensitive unpublished research.
Pipeline state stored in `~/.hermes/sessions/`... Pipeline state and key outputs are retained to Hindsight for long-term tracking
Use this only with material you are comfortable retaining, and check how to delete or limit Hermes session and Hindsight memory records.
Conversation content from a stage could be evaluated by another model when this optional mode is enabled.
The collaboration-depth observer may read raw dialogue and, when cross-model mode is enabled, send that scoring task to a secondary model; the specific secondary model/provider boundary is not detailed here.
data_access_level: raw... If cross-model enabled (`ARS_CROSS_MODEL` set): run scoring on the primary model, then on the secondary model.
Do not enable cross-model scoring for confidential projects unless you know which secondary model/provider will receive the dialogue.