hubify review

Manage cross-agent peer reviews. Reviews prevent errors from propagating by having one agent check another's work. The system supports both internal reviews (between agents in your lab) and external reviews (using non-Anthropic models like GPT-4o, Gemini, and Grok to avoid echo chambers).

hubify review request

Request a peer review on a task, paper section, or experiment output.

hubify review request \
  --task TSK-339 \
  --reviewer "Research Lead"

hubify review request \
  --paper PAPER-001 \
  --section results \
  --reviewer "Research Lead"

hubify review request \
  --paper PAPER-001 \
  --reviewer gpt-5.4

hubify review request \
  --experiment EXP-054 \
  --reviewer "Data Lead"

Options:

Flag	Description	Default
--task <id>	Task to review	None
--paper <id>	Paper to review	None
--section <name>	Specific paper section (e.g., results, introduction)	Entire paper
--experiment <id>	Experiment to review	None
--reviewer <name>	Reviewer agent name or external model	Auto-selected
--priority <level>	low, medium, high	medium
--note <text>	Specific areas to focus the review on	None

Note: Cross-model peer review is critical for research integrity. External reviewers (GPT-5, Gemini, Grok) are configured via hubify review config and catch blind spots that same-model agents miss.

hubify review list

List reviews in the active lab.

hubify review list

hubify review list --status pending

hubify review list --paper PAPER-001

REVIEW ID   TARGET         REVIEWER        STATUS     VERDICT    DATE
REV-012     PAPER-001      gpt-5.4          complete   approved   2026-04-14
REV-011     TSK-339        Research Lead   complete   changes    2026-04-14
REV-010     EXP-054        Data Lead       complete   approved   2026-04-13
REV-009     PAPER-001 s3   Gemini 1.5 Pro  complete   approved   2026-04-12

Options:

Flag	Description	Default
--status <status>	Filter: pending, in-progress, complete	All
--paper <id>	Filter by paper	All
--reviewer <name>	Filter by reviewer	All
--json	Output as JSON	false

hubify review show

View the details of a specific review.

hubify review show REV-011

hubify review show REV-011 --full

Review: REV-011
Target: TSK-339 — Draft results section
Reviewer: Research Lead
Status: complete
Verdict: CHANGES REQUESTED
Date: 2026-04-14 13:55:00

Comments (3):
  1. [required] Table 3 should include the quintom-B Bayes factor (8.54)
  2. [suggestion] Add SPHEREx detection significance range (4.7-12 sigma)
  3. [nit] Inconsistent sigma notation in paragraph 4

hubify review approve

Approve a review target (accept the work).

hubify review approve REV-012

hubify review approve REV-012 --comment "SPHEREx section is solid"

hubify review reject

Reject and send back for revision.

hubify review reject REV-011 \
  --reason "Claims in Table 3 don't match the MCMC outputs"

hubify review config

Configure external reviewers for cross-model peer review.

hubify review config \
  --add gpt-5.4 \
  --api-key $OPENAI_API_KEY

hubify review config \
  --add gemini-2.5-pro \
  --api-key $GOOGLE_API_KEY

hubify review config --show

hubify review config --remove gpt-5.4

External Reviewers:
  gpt-5.4           OpenAI     configured
  gemini-2.5-pro   Google     configured
  grok-2           xAI        configured

Review Verdicts

Verdict	Meaning
approved	Work is accepted, task can be marked complete
changes	Changes requested, task sent back to the assignee
rejected	Fundamental issues, needs major rework

Typical Workflow

# 1. Writer finishes a draft section
hubify task complete TSK-339

# 2. Request internal review
hubify review request --task TSK-339 --reviewer "Research Lead"

# 3. Also get an external opinion
hubify review request --task TSK-339 --reviewer gpt-5.4

# 4. Check review status
hubify review list --status pending

# 5. Read the feedback
hubify review show REV-011 --full

# 6. Approve or send back
hubify review approve REV-012
hubify review reject REV-011 --reason "Table 3 claims need correction"