advisor-update-writer
Advisor Update Writer
Write concise research updates that help an advisor, mentor, collaborator, or lab audience make decisions.
Use this skill when:
- the user needs a weekly update, advisor email, lab update, meeting memo, or collaborator status note
- experiment results, writing progress, reviewer risks, or implementation blockers need to be summarized
- the update should ask for a decision, feedback, resources, compute, paper-positioning advice, or priority choice
- project memory should be turned into a human-readable progress report
- several feedback loops need to be reconciled: algorithm, experiments, writing, review, rebuttal, artifact, or release
Do not use this skill for a full experiment report. Use experiment-report-writer when the main artifact is a detailed technical report. Use this skill when the main artifact is a decision-oriented communication.
Pair this skill with:
research-project-memoryto recover current status, decisions, claims, evidence, risks, and actionsexperiment-report-writerwhen detailed experiment evidence needs a source reportfigure-results-reviewwhen plots or tables will be shown to an advisorresult-diagnosiswhen negative or ambiguous results need a decisionpaper-positioning-plannerwhen the update asks for paper-story or target-venue feedbackpaper-evidence-boardwhen writing and experiment gaps need a synchronized action listwork-timeline-plannerwhen the update needs retrospective timeline evidence
Skill Directory Layout
<installed-skill-dir>/
├── SKILL.md
└── references/
├── decision-framing.md
├── memory-writeback.md
└── update-template.md
Progressive Loading
- Always read
references/decision-framing.mdandreferences/update-template.mdbefore drafting a saved or high-stakes update. - Read
references/memory-writeback.mdwhen the update creates decisions, actions, or durable feedback that should persist.
Core Principles
- Start with the decision-relevant delta since the last update.
- Separate facts, interpretation, risks, and asks.
- Do not bury the request. Make the advisor's needed input explicit.
- Compress raw experiment detail into evidence and implications; link to detailed reports when needed.
- Include negative results when they change the project decision.
- Turn feedback into actions and memory updates after the meeting or response.
- Match the audience: advisor updates are direct and decision-heavy; lab updates need more context; collaborator notes need ownership and handoff.
Step 1 - Classify the Update
Choose the mode:
weekly: progress since last update, current blockers, next week plandecision: options, evidence, recommendation, explicit askmeeting: agenda, discussion points, decisions needed, notes after meetingemail: concise message with context, status, ask, and attachmentslab: broader context, key result, what the audience should remember
Identify:
- audience and expected length
- deadline or meeting time
- project phase
- key evidence since last update
- unresolved decisions
- desired advisor action
- save path, if a file is requested
If saving and no path is given, use:
docs/updates/advisor_update_YYYY-MM-DD.md
Step 2 - Gather Current State
Prefer project memory when available, then primary files.
Look for:
memory/current-status.mdmemory/decision-log.mdmemory/claim-board.mdmemory/evidence-board.mdmemory/risk-board.mdmemory/action-board.md- experiment reports, paper drafts, reviewer notes, rebuttal notes, and recent git commits
Extract:
- what changed
- what evidence supports the change
- what failed or remains uncertain
- what decision is now required
- what the user recommends
- what happens next
Step 3 - Frame the Decision
Read references/decision-framing.md.
For each decision, produce:
- decision question
- options
- evidence for and against each option
- risks if delayed
- recommended option
- exact advisor ask
- next action after a yes/no/alternative answer
If there is no decision needed, frame the update around progress, risks, and next actions.
Step 4 - Draft the Update
Read references/update-template.md.
Default structure:
- one-line status
- key progress
- evidence and interpretation
- blockers or risks
- decision/ask
- next actions
For email, make the first paragraph enough to answer "what do you need from me?"
For meeting notes, include both agenda and post-meeting decisions if the meeting has already happened.
Step 5 - Check the Update
Before finalizing:
- every important claim has evidence or is marked as interpretation
- the ask is explicit
- blockers have owners or proposed next steps
- negative results are not hidden if they affect direction
- the update is short enough for the audience
- attachments or links are named
- any decisions or feedback are ready for memory writeback
Step 6 - Write Back After Feedback
Read references/memory-writeback.md when memory exists.
Update decisions, actions, risks, and status after the advisor responds or after meeting notes are finalized.
More from a-green-hand-jack/ml-research-skills
project-init
Initialize an ML research project control root. Use for paper/code/slides repos, shared memory, GitHub Project alignment, agent guidance, worktree policy, and lifecycle handoffs.
37project-sync
Sync verified code-side experiment results into paper memory. Use when logs, reports, run docs, or user-confirmed metrics should become paper-facing evidence.
36add-git-tag
Create annotated Git milestone tags. Use when completing a phase, releasing a version, or marking a research checkpoint.
36update-docs
Refresh project documentation after code changes. Use after implementing features, changing behavior, or preparing a milestone commit.
36init-latex-project
Initialize a LaTeX academic paper project. Use for new conference or journal papers needing templates, macros, venue preambles, and writing guidance.
36new-workspace
Create Git branches or worktrees for research code and paper versions. Use for experiments, baselines, rebuttal fixes, arXiv/camera-ready branches, and worktree memory.
36