Skill Claude CodeCodex
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me" or "grill-me".
Skill Claude CodeCodex
Set up and run the botference document-review interface (Google-Docs-style comments on rendered LaTeX/Markdown, bots replying in threads, apply/commit of accepted changes) in any repo — and follow the round protocol when the user asks for a review round or tags you from the review page.
Skill Claude CodeCodex
Applies to every reply, always — plain words, honest uncertainty.
Skill Claude CodeCodex
Render any document the user needs to review (design docs, plans, story bibles, proposals, reports) as a self-contained HTML page with Google-Docs-style commenting — highlight text, attach notes, export feedback. Use whenever producing or updating a document whose primary purpose is for the user to READ AND REACT TO…
Skill Claude CodeCodex
Look at any web UI you build or change — render it headless with botference see and read the screenshots — because layout and design failures produce NO errors or warnings. Use after every UI change, before declaring done, and whenever a human says something "looks wrong".
Skill Claude CodeCodex
Ideate and critique data visualizations using Edward Tufte's principles from "The Visual Display of Quantitative Information." Use this skill when: (1) Designing new data visualizations or charts (2) Critiquing or improving existing visualizations (3) Reviewing dashboards or reports for graphical integrity (4)…
Agent Claude Code
Fourteen agents forming the Botference loop. Each reads checkpoint.md, produces outputs, updates checkpoint, and yields. All agents inherit shared protocol from agent-base.md.
Agent Claude Code
All agents inherit these rules. Agent-specific prompts override when they conflict.
Agent Claude Code
Agent "agent-template" from angadhn/botference, covering identity, inputs (read these), tools, operational guardrails and output format.
Agent Claude Code
Coder — reads, modifies, and tests application code. Operates on the project's source files directly (not AI-generated-outputs). Produces working, tested code changes committed to the repo.
Agent Claude Code
Coherence reviewer — checks whole-manuscript consistency after editing. Produces a diagnostic report (read-only on .tex files). Four checks in order.
Agent Claude Code
Critic — structural assessor and quality gatekeeper. Five modes (from checkpoint's Next Task).
Agent Claude Code
Deep reader — reads PDFs to extract quantitative findings, contradictions, and synthesis material. Prioritizes high-value sections (results, methods) over front matter. Final iteration produces sectionmap.md bridging reading to writing.
Agent Claude Code
Editor — makes substantiated improvements to manuscript sections. Every edit justified by evidence (ledger entries, venue requirements, or explicit reasoning). Produces git-diffable changes to .tex files. Improves clarity, precision, and compliance within existing structure. One section per iteration.
Agent Claude Code
Figure stylist — reviews generated figures for visual clarity, readability, and publication standards. Reviews design quality only, not scientific content.
Agent Claude Code
You are a dispatch orchestrator for botference's parallel execution system. You run at phase boundaries to decide HOW to execute the next batch of work. You do NOT do the work yourself — you produce a JSON dispatch instruction that the bash loop executes.
Agent Claude Code
Paper writer — writes or revises journal paper sections. Three modes (from checkpoint's Next Task).
Agent Claude Code
You are the planning agent for botference. You guide the user through project setup and produce an implementation plan that build mode executes.
Agent Claude Code
Provocateur — finds gaps, blind spots, and unexplored angles via three lenses.
Agent Claude Code
Refactorer — restructures existing code to reduce duplication and improve clarity without changing behavior. Produces refactored source files with verified results.
Agent Claude Code
Research coder — writes and runs code for simulations, data analysis, and figure generation. Three modes.
Agent Claude Code
Role Analyst — reads job postings and candidate CV, produces a fit assessment and resume strategy for job applications.
Agent Claude Code
Literature scout — searches for and evaluates papers relevant to the project. Two modes (from checkpoint's Next Task).
Agent Claude Code
Security auditor — performs read-only security analysis of a specific domain (secrets, database, auth, client, dependencies, CI/CD, or encryption) and produces a structured findings report with evidence-based severity ratings.