agentleFS
mhxiemhxie/Atelier/skills/review/SKILL.md

review

Skill5 starsChanged 35 days agoMITView on GitHub

Goal review for quarterly full reviews and monthly light pulses.

192 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use close-spec, design-with-images and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • dzhngSkill

    close-spec

    When a spec under specs/<feature>/ is done shipping (a write-spec build, or any planned task that produced a spec), archive it to specs/done/ and rewrite it from a build-plan into a durable rationale — the why, the principles, the invariants — pointing back to the real code for the how. Use when implementation has landed and the plan no longer matches what shipped, or the user says a feature/spec is finished. Pairs with [write-spec](../write-spec/SKILL.md) (the plan this closes) and [review](../review/SKILL.md) (run before closing).

    Also in 5 projects that use this

  • dzhngSkill

    design-with-images

    Explore UI and visual design with image generation, then iterate the real implementation against the user-selected concept. Use when brainstorming visual directions, offering mockup variants, or implementing an approved generated design; do not stop at the mockups or the first code approximation.

    Also in 5 projects that use this

  • LambenthanSkill

    daily-arxiv

    每日从 arXiv 拉取新论文,过滤相关性,auto-ingest 高优先级论文,检测 SOTA 更新

    Also in 5 projects that use this

  • LambenthanSkill

    exp-eval

    实验判决门:Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges

    Also in 5 projects that use this

  • +3 more from dzhng
  • +5 more from Lambenthan

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 6 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 157 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If <30 days since last pulse → no pulse needed yet ("monthly pulse already done N days ago").

    line 35
  2. Distinctive:

    Use the artifact's latest session sections as the continuity seed.

    line 52
  3. Distinctive:

    Add today's daily note only if current capture is necessary for a disputed or missing goal state.

    line 59
  4. Distinctive:

    Read key goal-note sections directly.

    line 62
  5. Distinctive:

    After discussing, present the draft and obtain approval before writing:

    line 97
  6. Distinctive:

    Pulse write gate: no material change means no pulse file.

    line 101
  7. Very specific:

    Keep a short summary, progressing/neglected/shifted findings under the actual direction headings, commitment evidence and state, and source …

    line 106
  8. Very specific:

    Run uv run scripts/session_log.py --type review --duration <minutes> and fill the compact Continuity, Anomalies, and Operations sections …

    line 114
  9. Very specific:

    If previous review files exist in <paths.reflections>/, compare:

    line 132

“Common” is measured: kept by at least 6 in 10 of the 157 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Goal Review

Also reachable via /hi <natural language> (e.g., /hi review my goals, /hi goal pulse, /hi check directions). See harness/intents.toml [intents.review] for the row's example phrases. Both paths execute this same procedure.

Review progress on near/mid/long-term goals. Surface what's progressing, what's neglected, and what has shifted.

Cadence

TypeFrequencyWhat
Full reviewQuarterly (start of Q1 / Q2 / Q3 / Q4, i.e., early Jan / Apr / Jul / Oct)Walk every goal in directions.md, decide keep / shift / drop. Retire >1y stale goals. Trend table vs. prior reviews.
Light pulseMonthly5-min check inside /weekly or standalone: any goal materially advanced? Any neglected? Any newly born? No file write required if nothing surfaced.
Annual resetYearly (Jan or birthday)Full rebuild anchored to <year>小目标. Touches directions.md directly. Often paired with /introspect.

Inspect the last full review and the last pulse-equivalent so the monthly cadence does not produce duplicate pulses within 30 days. A pulse-equivalent is either a standalone *-review-pulse.md or a *-weekly.md, whose goal and commitment check supplies the monthly pulse. Take whichever is most recent:

Bash: last_full=$(find "$OV/reflections" -name '*-review.md' ! -name '*-review-pulse.md' 2>/dev/null | sort | tail -1)
Bash: last_pulse=$(find "$OV/reflections" \( -name '*-review-pulse.md' -o -name '*-weekly.md' \) 2>/dev/null | sort | tail -1)

Note: a no-change pulse intentionally skips writing a file (per Output → Pulse write gate), so last_pulse may understate by up to 30 days. That's tolerable: the cost of suggesting one extra pulse during the gap is far lower than the cost of a missed pulse.

Decision rules align with the cadence table above (full = quarterly = 90 days, light pulse = monthly = 30 days). Branch on last_full first, then refine with last_pulse:

  • No prior full review → run the full quarterly form to establish the baseline. Lookback = 90 days for Context Loading. First-run case.
  • <30 days since last full review → no review needed; tell user "last full review was N days ago, full quarterly cadence not due yet, no monthly pulse needed yet either". Offer full only if user explicitly asks.
  • 30-89 days since last full review → check last_pulse:
    • If <30 days since last pulse → no pulse needed yet ("monthly pulse already done N days ago"). Offer full only if user explicitly asks.
    • Otherwise (no pulse or ≥30 days) → light pulse default. The full quarterly cadence has not yet rolled over; offer full only if user explicitly asks.
  • 90-120 days since last full review (on cadence) → full quarterly form. The canonical quarterly review.
  • >120 days since last full review (overdue) → full quarterly form with an honest gap note ("last full review was N days ago, X days past the quarterly cadence").

Stale-goal floor: every full review must explicitly resolve each item under ## Goals requiring an explicit decision in directions.md (retire, redefine, or re-commit) and any goal older than 1 year with no progress evidence. Do not let stale goals roll over silently.

Prerequisites

The routed context helper below stops toward /introspect for a missing declared profile and reports staleness for selected profiles.

Context Loading

  1. Reuse the current review Repomix context artifact from $hi; for direct invocation, run uv run scripts/context_bundle.py --intent review.
  2. Use the artifact's latest session sections as the continuity seed. Search the selected lookback window, 90 days for a full review or 30 days for a pulse, and triage QMD candidates before reading 3 to 5 relevant source sections. Do not preload every reflection in the window.
  3. Pull goal-related updates from the local vault, bounded to the lookback window. Do NOT issue an unbounded Grep(path: "$OV/") — an unbounded grep will pull stale historical matches that skew the review. Use find -print0 | xargs -0 grep so recency actually binds. Substitute <N> with the lookback (90 for full, 30 for pulse):
    • Bash: find "$OV"/daily "$OV"/reflections "$OV"/gtd "$OV"/wiki -type f -name "*.md" -mtime -<N> -print0 2>/dev/null | xargs -0 grep -HnE "目标|goal|progress|进展|milestone" 2>/dev/null: recency-bounded goal and progress mentions across both languages in one pass. Safe with an empty working set (xargs does nothing if stdin is empty).
    • Add today's daily note only if current capture is necessary for a disputed or missing goal state.
  4. Read key goal-note sections directly. Read a complete file only when the full goal record is required. If a referenced note is genuinely missing from the local vault, report the gap.

Analysis

Walk directions.md in its current shape: one pass per H3 under ## Mid-term direction, then every row of the commitments table (full review) or the rows due since the last pulse (monthly), then ## Long-term direction on full reviews only: does each mid-term heading still ladder into a long-term line, and does any long-term line have no mid-term carrier? The category names come from the file at runtime, never from this page.

Progressing

Which goals have evidence of recent activity? Look for:

  • Goal notes edited recently
  • Daily notes mentioning goal-related topics
  • Reflections that referenced these goals

Neglected

Which goals appear in directions.md but have NO recent activity? Look for:

  • Goals not mentioned in any recent reflection
  • Goal notes not edited in 30+ days
  • Topics absent from daily notes

Shifted

Have the user's priorities changed? Look for:

  • New topics appearing frequently in recent notes that aren't in directions.md
  • Goals that were active 3 months ago but have gone quiet
  • Contradictions between stated goals and recent behavior

Review Output

Present the review interactively, category by category. For each finding, cite the specific source note.

After discussing, present the draft and obtain approval before writing:

File: <paths.reflections>/YYYY-MM-DD-review.md for full reviews; <paths.reflections>/YYYY-MM-DD-review-pulse.md for monthly light pulse runs. The distinct suffix lets the Cadence Bash check (above) tell them apart so a pulse does not silently defer the next quarterly review.

Pulse write gate: no material change means no pulse file. Full reviews still produce a draft when goals are unchanged, so an approved save can anchor the quarterly cadence. A declined save must not be reported as a completed cadence anchor. Neither branch changes directions.md without separate approval.

Keep a short summary, progressing/neglected/shifted findings under the actual direction headings, commitment evidence and state, and source links. Full reviews also show the long-term ladder and explicit resolutions of stale goals. Include new interests, experiments, or a trend comparison only when supported; do not fill fixed category templates or manufacture a quota of actions.

Session Log

Run uv run scripts/session_log.py --type review --duration <minutes> and fill the compact Continuity, Anomalies, and Operations sections per protocols/session-log.md. Record the next review trigger and actual save/update outcomes, including a no-change pulse or declined write. Logs never block the session.

Wrap Up

The approved review file is the durable output. Daily notes remain read-only. Report the actual saved location, no-change pulse, or declined write; never announce a save that did not happen.

Suggest follow-ups:

  • /hi for daily check-ins between reviews
  • /weekly for lighter weekly check-ins
  • /decision if any goal needs a decision about continuing or stopping

Trend Analysis (if prior reviews exist)

If previous review files exist in <paths.reflections>/, compare:

  • Which goals were "Neglected" last time — are they still neglected? (Chronic neglect signal)
  • Which goals were "Progressing" — have they continued? (Momentum signal)
  • Which "Suggested Experiments" from last review were actually done? (Follow-through signal)

Present trend as:

GoalLast ReviewThis ReviewTrend
[goal]ProgressingProgressingSustained momentum
[goal]NeglectedNeglectedChronic neglect — needs decision
[goal]ProgressingNeglectedLost momentum — investigate

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.